The influence of the oil price on the use of subjectivity in CEO performance-based compensation in the U.S. Oil & Gas industry

The influence of the oil price on the use of subjectivity in CEO

performance-based compensation in the U.S. Oil & Gas industry

Name: Eric Mars

Student number: 10091998

Thesis supervisor: Mr. M. Schabus Date: June 26th_{, 2017}

Word Count: 24,336

MSc Accountancy & Control - Accountancy and Control specializations Faculty of Economics and Business, University of Amsterdam

Statement of Originality

This document is written by student Eric Mars 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.

Abstract

This paper investigates the oil price influence on the use of subjectivity in CEO performance-based compensation in the U.S. oil and gas (hereafter O&G) industry from 2000 until 2015. The study uses a sample of O&G and non-O&G companies to compare the level of subjectivity use. The analysis consists of two stages, using robust regressions. The first stage regresses granted executive compensation on observable financial company performance, with the residual of this regression being a proxy for the level of subjectivity used by the BoD. The second stage analyses regress this proxy on, amongst others, an O&G industry dummy and the annual oil price change. Results indicate that executive performance-based compensation in the O&G industry has a larger subjective component. However, no evidence is found on the oil price causing this increased used of subjectivity. Thus although research argues the oil price to be a primary determinant of O&G industry performance (Boyer and Filion, 2007 ; Berry and Wright, 2001), overall the hypothesized association between the ex-post use of subjectivity to reduce noise in performance measures (Gibbs et al., 2004) and the oil price cannot be supported with empirical evidence.

Table of Contents

Section

Page

2.2.1 Compensation package components ... 7

2.2.2 Performance-contingent (formula) bonus ... 8

2.2.3 Objective and subjective performance measures ... 9

2.2.4 Shortcomings of objective performance measures in performance evaluation... 10

2.2.5 Relative performance evaluation ... 11

2.2.6 The use of subjectivity in performance evaluation and compensation ... 12

2.2.6.1 Types of subjectivity and their effects ... 12

2.2.6.2 Shortcomings of subjectivity ... 13

2.2.6.3 Previous research on subjectivity use and uncontrollable factors ... 14

2.3 The oil price impact on firms in the O&G industry ... 15

2.3.1 O&G industry structure and activities ... 15

2.3.2 Oil price and profits in the O&G industry ... 16

2.3.3 Oil price determinants ... 17

2.4 Hypotheses ... 20

3. Methodology ... 25

3.1 Sample selection ... 25

3.2 Empirical method ... 28

3.3 Control variables: Determinants of the use of subjectivity ... 32

4. Results and discussion... 34

4.1 Univariate statistics ... 34

4.1.1 Summary statistics... 34

4.1.2 Pearson pairwise correlation matrix ... 36

4.2 Multivariate analysis ... 37

4.2.1 Stage 1 ... 39

4.2.2 Stage 2 ... 40

4.3 Robustness tests ... 45

4.3.2 Executive compensation ... 47

4.3.2.1 Total bonus (TBONUS) ... 49

4.3.2.2 Formula bonus (FBONUS) ... 52

4.3.2.3 Discretionary bonus (DBONUS) ... 54

4.3.3 First stage comprehensive observable performance (COMP. FIRST STAGE) ... 55

4.4 Comparing conclusions from main analysis and robustness analyses ... 59

5. Conclusion ... 60

6. References ... 63

1

1. Introduction

The problems listed companies encounter as a result of the separation of ownership and control are not new. Since the early 1900s large listed companies have become more commonplace in the business landscape at the expense of the typical owner-managed and family businesses, making these problems ever more prominent (Murphy, 2012). These so-called agency problems emerge due to a conflict of interests between the company’s executives, who are delegated decision-making authority to control the company and steer company activities (on behalf of the owners), and the company’s owners (Jensen and Meckling, 1976). The main organizational objective for listed companies is to maximize shareholder value. However, executives with divergent interests could make decisions that are favorable for their compensation/private wealth, but which do not maximize firm value. As such, these executives are not fully acting in the owners’ interests (Merchant and van der Stede, 2012).

The Board of Directors (hereafter BoD), acting on behalf of the shareholders, aims to reduce this conflict and other agency costs by including performance-based compensation as a component of the executive’s remuneration package (Murphy, 2012). This compensation is contingent on ex-ante specified performance measures and executives are therefore incentivized to direct their efforts towards the achievement of organizational objectives (Merchant and van der Stede, 2012). Appropriate performance measures reflect the organizational objectives that have high priority and thereby realign the executive’s interests with those of shareholders. Objective quantitative performance measures are favored because of their invulnerability to bias, such that the individual’s performance evaluation can be verified by a third party (Merchant and van der Stede, 2012).

However, previous research has highlighted several weaknesses in the use of objective performance measures. Solely using objective performance measures in incentive contracts for performance-based compensation tends to be insufficient in inducing behavior that is congruent with organizational objectives (Gibbs et al., 2004). Furthermore, research has pointed out that objective performance measures are often subject to gaming, manipulation and noise and therefore do not capture the subordinate’s true contribution to organizational performance (Merchant and van der Stede, 2012).

Building on the noise related concern, compensation that is based on objective performance measures could lead to compensation being driven mainly by external factors rather than the decisions and exerted effort of the executive. Previous research has shown that subjectivity can

2

be used to address the shortcomings of using solely objective performance measures in compensation structures. In doing so, executives continue to be properly incentivized to display desirable behavior that is in alignment with organizational objectives (Bushman et al., 1996 ; Gibbs et al., 2004 ; Merchant and van der Stede, 2012).

A setting in which subjectivity use could be more prevalent is one where external uncontrollable factors have a substantial influence on the executive’s performance (Merchant and van der Stede, 2012). These factors introduce noise and result in the performance measures not adequately capturing the executive’s actual contribution to the company’s performance, thereby reducing the effectiveness of objective performance measures (Gibbs et al., 2004). The Oil & Gas (hereafter O&G) industry could be an interesting setting to investigate whether this notion holds. In the O&G industry company profits and share prices are substantially impacted by the oil price level. Amongst others, Boyer and Filion (2007), Elyasiana et al. (2011), Ramos and Veiga (2011) and Dayanandan and Donker (2011) have argued that O&G companies’ profitability and stock prices are positively correlated with the oil price level. One does not have to dig in to academic research to observe this phenomenon, as analysts and general media also thoroughly cover the unfolding of significant events in this industry (Forbes, 2016). With the price of e.g. Brent crude oil has dropping from approximately $108 per barrel in July 2014 to $51.86 per barrel in October 2016, the oil price level has dropped by more than 50% (Nasdaq, 2016). This oil price collapse has led to a substantial decrease in O&G companies’ profitability, value and subsequent stock prices (Forbes, 2016).

The implication of these findings is that company performance is heavily impacted by a factor on which company executives have no control. Oil is a globally traded commodity and the price fluctuations are due to, amongst others, the global supply and demand for oil and policy changes of intergovernmental organizations such as the Organization of the Petroleum Exporting Countries (OPEC) (Smith, 2009 ; Bertrand and Mullainathan, 2001). This makes it unlikely that the CEO of a single O&G company can have any influence on the oil price level. However, the measures on which the executive’s performance are evaluated when determining the amount of performance-based compensation to be granted are primarily based on accounting and stock market measures, which are substantially affected by this external factor (Gibbs et al., 2004 ; Bol, 2008). As such, fluctuations in these performance measures do not reflect the true contribution of the executive’s actions to firm performance, which reduces the effectiveness of using these measures to incentivize subordinates (Merchant & van der Stede, 2012 ; Gibbs et al., 2004).

3

The use of subjectivity can complement objective performance measures by correcting an executive’s perceived performance for the noise of this uncontrollable factor (Murphy, 2012 ; Merchant and van der Stede, 2012). However most studies until now have found subjectivity primarily being used to make upward adjustments when noise has a negative impact on performance measures, rather than to make downward corrections when noise has a positive impact too (Gibbs et al., 2004 ; Bol and Smith, 2011 ; Merchant and van der Stede, 2012). In this situation subjectivity is used to compensate employees even though e.g. targets were not achieved, by reducing the influence of external factors that have a negative impact on performance measures. The purpose of this is to increase the ‘fairness’ of the performance evaluation and to reduce the risk subordinates face by reducing the impact of factors which the subordinate has little control over. If the BoD didn’t have this option to evaluate employees in hindsight, employees would demand a risk premium on top of the usual salary to compensate for the noise this external factor is creating (Merchant and van der Stede, 2012).

Building on the previous, this study investigates whether the use of subjectivity is more prevalent in an industry where performance is heavily impacted by an external factor in comparison to an industry where such an impact would be less obvious. This will be done by comparing subjectivity use in the O&G industry to a control group of other industries. Another aim of this study is to determine whether the general finding of prior research also holds in the O&G industry. That is, whether the use of subjectivity is also asymmetric in this industry. Due to the oil price being such an important factor driving O&G companies’ profits and share prices, one would expect the use of subjectivity to not be uncommon in this industry. If performance-based compensation would be mainly based on objective measures, CEOs would be harshly punished in times of depressed oil prices while being generously rewarded during upward oil price trends. In other words, the granted performance-based compensation would not truly reflect the outcome of the executive’s decisions and exerted effort. This lack of executive stimulation could lead CEOs to show behavior that is deviant from organizational objectives.

These research opportunities are summarized in the following broad research question, which will be applied to the O&G industry in the United States from 2000 until 2015.

 How is the use of subjectivity in CEO performance-based compensation in the O&G industry influenced by the oil price?

Previous research on the application of subjectivity in performance evaluation and compensation has mostly focused on lower-level employees. In this setting the subordinate’s tasks can be easily

4

defined and anticipated, making it fairly easy to measure the subordinate’s performance objectively (Gibbs et al., 2004 ; Bol, 2008 ; Merchant and van der Stede, 2012). Fewer studies have focused on the use of subjectivity in performance-based compensation at the executive level. This is because a CEO’s tasks are subject to a lot of discretion, are difficult to define and anticipate (i.e. have low programmability), and observing the executive’s exerted effort is more problematic (Grabke-Rundell and Gomez-Mejia, 2002). Furthermore, one important finding of research on subjectivity is its use to protect subordinates from downside risk, whilst it being less prevalent to correct performance influenced by positive effects of the uncontrollable factors on the upside (Gibbs et al., 2004 ; Höppe and Moers, 2011 ; Bol and Smith, 2011). This asymmetric use of subjectivity could be less likely to exist in an industry in which an external factor is of such influence on profits and market valuation. Findings that support this suggestion in the O&G industry would question the transferability of conclusions drawn from previous academic research and therefore provide a new insight on the use of subjectivity.

The remainder of this paper will proceed as follows. Section 2 will give an overview of the relevant literature. Section 3 covers the data sample and the empirical method used. The results of the analysis are presented and discussed in section 4, while section 5 will provide the conclusion of this study.

5

2. Literature review

2.1 Agency theory

Although the negative organizational impacts had been observed for years, it was not until the 1960s that progress in research on these phenomena really began to pick off. A well-received article by Jensen and Meckling (1976) has been argued to be the first to apply agency theory to a company setting. This research describes the relationship between the principal (the BoD on behalf of the shareholders) and the risk-averse agent (the company’s CEO) and its consequences and organizational impacts.

The agent (CEO) is appointed by the principal (BoD) to conduct business such that the shareholders’ interests and objectives are honored. The main organizational objective for listed companies is the maximization of shareholder value. As such, the principal is charged with overseeing the agent’s activities to ensure that they contribute to maximizing shareholder value. The principal attempts to do this by monitoring the agent’s activities to assess the agent’s performance (Merchant and van der Stede, 2012).

Issues arising within the principal-agent relationship originate from two essential circumstances: the presence of information asymmetry (where one party has superior information over the other party) and the presence of conflicts of interest between the agent and the principal. For these two factors to exist in a relationship, agency theory necessitates agents and principals to have divergent interests, the agent to perceive his exerted effort as being costly, and the agent to be risk-averse and self-interested (Eisenhardt, 1989).

Broadly speaking, contracting issues in the principal-agent relationship are categorized as either adverse selection or moral hazard problems. The essential difference between the two is the timing of the information asymmetry occurrence; before (ex-ante) or after (ex-post) a contract is established between the principal and the agent. With adverse selection problems information asymmetry arises before a contract is established between the principal and the agent. The information asymmetry relates to the characteristics of the agent, which are not perfectly observable by the principal. As such, the principal is unsure whether the agent is capable of fulfilling the CEO function and the principal cannot distinguish more capable agents from agents who are not fit for the job. However, adverse selection problems are less relevant for the empirical research proposed in this study. The issues in the principal-agent relationship related to this research belong to the other category: moral hazard problems.

6

Unlike adverse selection problems, moral hazard problems occur after the contract has been established and once the agent starts to carry out tasks related to his job function. As such, information asymmetry arises when the principal has difficulty monitoring the agent’s made decisions and their consequences for the organization. Monitoring difficulties are especially prevalent when the agent is a CEO, as the typical CEO function and its activities are characterized by low programmability, high discretion and low measurability (Rajan and Reichelstein, 2006). When there are divergent interests, the agent prioritizes his own personal interests over those of the shareholders and therefore the agent’s actions do not contribute to the maximization of shareholder value. The difficulty of monitoring the consequences of the agent’s actions makes the principal less informed than the agent as to the agent’s exerted effort, the reasons behind decisions and conducted activities, and what the consequences of these actions are for the organization (Jensen and Meckling, 1976). As such, the executive could be tempted to e.g. make sub-optimal investment decisions based on unobservable personal interests, shirk on the job and show behavior that is deviant from maximizing shareholder value. In response to this situation the principal attempts to steer the agent’s effort towards the achievement of organizational objectives. One approach to do this is to design the agent’s compensation structure in contracts to reflect the desired behavior. If the compensation structure is appropriate, the increase in effort exerted by the agent should progress the organization towards the fulfillment of its objectives (Murphy, 2012).

Agency costs attributed to the principal-agent relationship arise in three forms: monitoring costs incurred by the principal to ensure agents are displaying congruent behavior, bonding costs related to compensating agents to increase the alignment of the agent’s interests with those of the principal, and lastly costs termed as ‘residual losses’ (Jensen and Meckling, 1976). A residual loss is the difference in organizational value generated by the decisions made by the agent and decisions that would maximize principal welfare (decisions that are fully congruent). The first two forms of agency costs mentioned are incurred to decrease the amount of residual loss. Thus organizations are continuously making tradeoffs between the costs of instruments inducing congruent agent behavior and the benefits of these instruments (being the reduction of the residual loss).

2.2 Executive compensation

As previously stated, a well-structured executive compensation package should increase the alignment of interests between the principal (the BoD on behalf of the shareholders) and the

7

agent (CEO). Consequently, the agent’s exerted effort to maximize his personal (monetary) wealth should also lead to an increase in firm value (Dechow and Sloan, 1991). In other words, the compensation package should induce goal congruence. The compensation package executives tend to be offered consists of a fixed base salary and an incentive component. The incentive component usually includes an annual performance-contingent monetary bonus and incentives focused on the long-term such as stock options or similar equity instruments (Bushman et al., 1996 ; Murphy, 2012). Additionally, the BoD also has the authority to grant a discretionary monetary bonus to the CEO in hindsight. The final component of a compensation package includes pension plans and extra perquisites such as insurance and car leases.

2.2.1 Compensation package components

A fixed base salary is necessary to attract talented and skilled (risk-averse) agents who are capable of carrying out the CEO function. It should therefore be in line with the market; the base salary level should be in close range to base salaries offered for similar functions in comparable companies to make talent interested in the job (Conyon, 2006).

Performance-contingent (formula) bonuses typically cover an annual period and are therefore categorized as short-term incentives (Merchant and van der Stede, 2012). The level of the bonus granted is dependent on how well the agent has performed, based on ex-ante specified performance measures. This component of the compensation package will be described in more detail in the next section.

Stock options and similar equity instruments such as restricted stocks are seen as long-term incentives as they typically have a more long-term focus and cover a period of multiple years. These instruments are used to increase the alignment of the CEO’s personal interests with those of the shareholder, who is interested in an increasing share price over the longer term (Merchant and van der Stede, 2012 ; Berk and Demarzo, 2013). Stock options give the executive the right to purchase a specified number of shares at a pre-determined price (strike price). As such, the executive’s effort is directed to increasing shareholder value, as the options can only be exercised once the share price has reached the strike price. Specific requirements attached to these stock options such as vesting schedules and forfeitures upon termination prevent executives from misusing this component (Merchant and van der Stede, 2012). For example, an executive could inflate the stock price, gain the proceeds and leave the company in the short-term (Merchant and van der Stede, 2012). Other equity instruments also have similar requirements to prevent misuse. For instance, restricted stock also directs the executive towards a long-term focus by prohibiting

8

the executive from selling the stock for a specified number of years. Companies sometimes make the granting of these instruments or the eligibility to exercise stock options contingent on performance measures in a similar way as performance-contingent bonuses.

Discretionary bonuses, unlike performance-contingent bonuses, are not awarded based on the outcomes of ex-ante specified performance measures (Ederhof, 2010). The executive is not aware of the amount, timing or requirements related to this bonus. Rather, the BoD grants a discretionary bonus to the executive ex-post for whatever reason the BoD finds necessary. Another component of executive compensation relates to perquisites such as the use of company cars or jets, housing accommodation and insurance coverage. Unlike the other components this category is hard to grasp due to the lack of reliable data, although in recent years regulators have been introducing stricter disclosure requirements.

2.2.2 Performance-contingent (formula) bonus

The performance-contingent bonus is an important instrument for the BoD to trigger CEO behavior that is in alignment with organizational objectives. Prior to the period of performance (ex-ante), the BoD specifies a number of performance measures and the target levels of the performance measures which the BoD would like to have realized. These specifics directly influence the amount of bonus that will be awarded. Upper and lower bounds are established around the set target to provide a minimum and maximum performance threshold, in between which the subordinate is entitled to a bonus amount (Murphy, 2012). After the period of performance (ex-post), the executive is evaluated on his performance by comparing the outcomes of the performance measures to the targets. The level to which the company has progressed on these performance measures (or the extent to which the targets have been achieved) reflects the executive’s performance and determines the amount of bonus to be awarded.

Rewarding subordinates according to their contribution to the achievement of organizational objectives generally increases subordinates’ performance, as subordinates recognize that exerting additional effort will be recognized and rewarded with a higher level of pay (Lazear, 2000 ; Prendergast, 1999 ; Merchant and van der Stede, 2012). However, there are also factors that contribute to increased subordinate performance by using this instrument. These include increased subordinate awareness and understanding of BoD expectations and the company’s strategy, directing subordinate effort towards the company’s critical success factors and

9

monitoring effects of periodically evaluating the impact of subordinates’ effort and decisions on company performance (Gibbs et al., 2004 ; Merchant and van der Stede, 2012).

2.2.3 Objective and subjective performance measures

When constructing a (formula) bonus based on performance measures, one needs to consider the effectiveness of these measures. That is, the selected performance measures combined should provide a complete, accurate, informative and timely picture of an individual’s contribution to the organization (Gibbs et al., 2004). Aspects of performance measures that determine their effectiveness are those of controllability (the extent to which the measure is sensitive to the individual’s efforts), susceptibility to uncontrollable factors (noise), congruence (the extent to which an increase in the performance measure leads to an increase in firm value) and their vulnerability to manipulation (Gibbs et al., 2004). In addition, one should be able to establish targets unambiguously prior to the period of the performance and subordinates should have an understanding of how their efforts translate into changes in the performance measures (Speklé and Verbeeten, 2014).

In determining which performance measures a subordinate’s evaluation process should be based on, Holmstrom (1979) presents the informativeness principle. When a subordinate’s actions are imperfectly observable (not all dimensions of the job are captured in the performance measure), any performance measure that provides additional information regarding the subordinate’s effort should be included in the subordinate’s evaluation too.

Performance measures are categorized as either being objective or subjective. Starting with the former, objective performance measures are quantitative in nature and are financial measures that reflect accounting-based or market-based performance (Merchant and van der Stede, 2012 ; Conyon, 2006 ; Dechow and Sloan, 1991). Examples of accounting-based measures are net profit, operating profit, ROI (return on investment) and ROA (return on assets). Market-based measures relate to changes in the company’s share price or shareholder returns. These financial performance measures are easy to observe. Furthermore, the evaluation process is unlikely to be questioned as financial accounting and reporting is an obliged component of listed companies’ activities and is subject to regulatory, investor and internal scrutiny. However, non-financial performance measures, although far less common, can also be objective (e.g. the number of machine breakdowns in a manufacturing company). Objective performance measures are favored for their verifiability. That is, a third party other than the superior and the subordinate

10

can assess whether the superior’s evaluation of a subordinate’s performance is accurate (Bol, 2008).

Subjective performance measures on the other hand are non-financial and are qualitative in nature, as well as being difficult to measure (e.g. leadership strength). These measures require the superior to make a judgment on how a subordinate has performed based on personal impressions and opinions instead of facts (Bol, 2008). The superior’s impressions could possibly emerge from a combination of results measures and circumstantial knowledge relevant to the subordinate’s activities (Merchant and van der Stede, 2012). Circumstantial knowledge can arise from e.g. direct observations and informal reports (Rajan and Reichelstein, 2006). Unlike objective measures, subjective measures are not verifiable. Thus for subjectivity to be effective, it is essential that superiors make fair and unbiased decisions and that subordinates accept these judgments and do not try to inappropriately influence the superior (Gibbs et al., 2004). With that being said, in certain job environments the application of subjectivity is crucial to keep subordinates incentivized.

2.2.4 Shortcomings of objective performance measures in performance evaluation

Although objective performance measures are favored in a subordinate’s performance evaluation, using solely objective performance measures in certain circumstances and for certain individuals can be ineffective or even harmful for the organization and the achievement of its objectives. The complexity and broad impacts of a job tend to increase the higher the job function in the company (Conyon, 2006).

Gibbs et al. (2004) point to the narrow focus subordinates display when evaluated based on certain performance measures. If the set of measures used is incomplete, in that it fails to cover all aspects of a subordinate’s job, the uncaptured aspects are more likely to be disregarded by the subordinate as they are not rewarded. This is especially apparent when performance measures are tied to accounting numbers. Financial accounting and reporting is conservative and tends to be backward-looking (Merchant and van der Stede, 2012). That is, accounting standards do not always recognize the future benefits of present decision-making, whilst the costs of these decisions are to be expensed immediately (e.g. Research & Development). Certain aspects of a company’s operations are disregarded altogether, such as internally generated goodwill. This stimulates subordinates to adopt a short-term focus. That is, subordinates will pursue activities that lead to immediate increases in the performance measures while neglecting activities that are essential for the long-term viability of the company but trigger a decrease in the performance

11

measures in the short-term (Dechow and Sloan, 1991). Performance measures based on market returns could also lead the executive to boost share prices in the short-term at the expense of long-term value generation.

Merchant and van der Stede (2012) also indicate that the vulnerability of objective performance measures to external factors reduces executive controllability on those measures. When companies use noisy performance measures, fluctuations in the performance measure directly affect the subordinate’s performance evaluation and compensation, even though the subordinate has had little to no influence on these fluctuations (Gibbs et al., 2004). This leads to an inaccurate assessment of the subordinate’s true contribution to the organization. But it also imposes a higher level of risk on the subordinate, who will possibly demand a risk premium. Objective performance measures are also prone to manipulation, especially when subordinates have a thorough understanding of how they are assessed (Baker et al., 1994 ; Merchant and van der Stede, 2012 ; Gibbs et al., 2004). Baker et al. (1994) explain how managers at a food processing company, whose bonuses were contingent on earnings growth, smoothed earnings to establish a consistent earnings growth over the years. This was achieved by manipulating revenues through altering the timing of shipments of goods to customers, as well as prematurely paying for future services that were to be provided by external parties. Gibbs et al. (2004) point out that subordinates have an informational advantage over their superiors regarding their activities and environment. Subordinates could use this advantage to inflate objective performance measures outcomes in their favor by using real-economic (e.g. finessing investment or pricing decisions) or accounting-based (accruals) manipulation techniques. Needless to say, these actions are not in line with shareholder interests. They do not contribute to maximizing firm value as they are solely undertaken to increase the subordinate’s individual wealth (Gibbs et al., 2009).

As already mentioned, a job’s complexity and organizational impact tend to increase the higher the job function in the company (Conyon, 2006 ; Merchant and van der Stede, 2012). As such, a subordinate’s effort is less likely to be captured by verifiable performance measures. Even though aggregate financial measures are typically more responsive to actions of executives than those of lower-level employees, certain aspects of the executive’s job remain difficult to quantify. 2.2.5 Relative Performance Evaluation

Regarding the influence of uncontrollable factors on objective performance measures, theory provides one possible solution. Relative performance evaluation (hereafter RPE) can reduce the

12

impact of uncontrollable factors using objective performance measures. RPE is achieved by selecting companies that are similar to the company in question, thereby forming a peer group to which executive performance can be compared (Merchant and van der Stede, 2012). Similarities which the peer companies are required to have include the company’s size, organizational structure, the way in which it is financed and the nature of its business activities. In doing so the uncontrollable factors which affect the company’s performance also affect the performance of the peer group. The executive’s performance can then be evaluated according to how well the company has performed in comparison to the peer group.

However, like other objective performance measures RPE also has its shortcomings. For instance, CEOs could try to persuade the BoD to select certain companies for the peer group. Most often, competitors with high performance are neglected whereas companies that are underperforming are favored (Gong et al., 2011). This increases the executive’s chances of outperforming the peer group. Moreover, selecting an appropriate peer group and monitoring its performance is costly (both time and effort wise). The findings of previous empirical research on whether RPE is actually effective are mixed (Albuquerque, 2009). Thus additional measures are required to effectively manage the impact of uncontrollable factors on executive performance. One such approach, the use of subjectivity, will be described in the next section.

2.2.6 The use of subjectivity in performance evaluation and compensation 2.2.6.1 Types of subjectivity and their effects

Subjectivity is included in the subordinate’s performance evaluation to address the shortcomings of objective performance measures, thereby increasing the effectiveness of performance-based compensation as an incentive (Baker et al., 1994 ; Gibbs et al., 2004 ; Merchant and van der Stede, 2012). The three main ways in which subjectivity is applied are: the use of subjective performance measures, the application of an ex-post subjective weighting to objective performance measures and the ex-post use of subjectivity to adjust compensation based on other factors than the ex-ante specified performance measures (Gibbs et al., 2004 ; Bol, 2008).

Subjective performance measures can complement objective performance measures in performance evaluations by capturing certain value-enhancing aspects of a subordinate’s job that are difficult to quantify (Gibbs et al., 2004). By incorporating these aspects, the performance measures put together provide a more comprehensive picture of a subordinate’s contribution to organizational objectives. As a result, subordinates are less likely to disregard these aspects as this will negatively affect their performance and subsequent compensation.

13

Applying a subjective weighting to performance measures or applying subjectivity to adjust compensation based on performance measures or other factors in hindsight allows for a more accurate evaluation of a subordinate’s performance. Not only can one reduce the impact of uncontrollable factors on subordinate compensation, one can also reduce the likelihood that subordinates will display non-productive behavior through manipulation (Gibbs et al., 2004). By providing this discretion to superiors, subordinates are more properly incentivized as compensation can be adjusted ex-post using information that became available during the period of performance. The superior can disregard manipulative actions, reduce the influence of uncontrollable factors or take the difficulty of achieving set targets into account when determining the amount of compensation a subordinate is entitled to (Merchant and van der Stede, 2012).

2.2.6.2 Shortcomings of subjectivity

However, subjectivity in performance evaluations also has its flaws. For subjectivity to be effective it is essential that superiors make fair and unbiased decisions and that subordinates accept these judgments and do not try to inappropriately influence the superior (Gibbs et al., 2004 ; Merchant and van der Stede, 2012).

Moreover, there are several aspects that can tarnish the effectiveness of subjective assessments. Since subjective assessments (both subjective performance measures and the application of subjectivity in performance evaluations) are not verifiable, superiors can disregard their pledges (Merchant and van der Stede, 2012). Subjectivity can also lead to inaccurate assessments of a subordinate’s performance when superiors do not make fair and unbiased judgments. Superiors tend to overrate underperformers to e.g. avoid personal confrontations with these individuals, at the expense of appropriately rewarding individuals with outstanding performance (Bol et al., 2016). The absence of fairness and unbiasedness in assessments can also lead to subordinates being judged based on their relationship with the superior, with subordinates who are more likeable being rewarded more generously than subordinates who do not have a good relationship with the superior. Furthermore, the uncertainty regarding measurement criteria or the lack of understanding of how one is evaluated in a subjective assessment is larger than with objective assessments, making it harder for the subordinate to decide which actions are appropriate (Merchant and van der Stede, 2012).

Put together, these flaws can lead to subordinates feeling frustrated and demotivated and can lead to a damaged working relationship between the superior and the subordinate.

14

Another cost of subjectivity relates to inappropriate and unproductive subordinate behavior aimed at influencing a superior. Subjective assessments are also expensive as superiors need to commit more time to inform themselves of the circumstances the subordinate was subject to, such that the superior can make well thought through decisions on how well a subordinate has performed (Merchant and van der Stede, 2012).

2.2.6.3 Previous research on subjectivity use and uncontrollable factors

Most prior research finds the use of subjectivity to reduce the impact of uncontrollable factors on a subordinate’s performance to be asymmetric (Merchant and van der Stede, 2012 ; Gibbs et al., 2004 ; Bertrand and Mullainathan, 2001 ; Bol and Smith, 2011). That is, superiors seem to apply more subjectivity when the uncontrollable factors have a negative impact on performance in comparison to when these factors have a positive impact on performance. Subjectivity is applied ex-post by overriding performance measure outcomes or by applying subjective weightings to performance measures, to protect the subordinate from missing compensation due to negative impacts on performance by factors that were out of the subordinate’s reach (Gibbs et al., 2004 ; Höppe and Moers, 2010). This reduces the risk subordinates face and reduces subordinates’ concerns of the performance evaluation being unfair, as the focus is placed on isolating and analyzing the subordinate’s true contribution to the firm and rewarding that contribution accordingly.

It is therefore surprising that most research does not find subjectivity to be applied to correct performance that is driven by positive impacts of uncontrollable factors. Superiors, responsible for ensuring that the owners’ interests are respected, should prevent or reduce the amount of rewards being granted to subordinates based on performance that was merely driven by these positive impacts and not by the subordinates contribution to organizational value (Merchant and van der Stede, 2012).

One possible explanation to this finding is the superior’s concern that a downward adjustment to performance is perceived by the subordinate to be a move to limit costs by increasing the performance standard in response to the performance being better than expected, thereby damaging the working relationship between the two (Gibbs et al., 2004). Another possible explanation is that of potential confrontation costs the superior endures when facing a disgruntled subordinate. The superior could choose to not follow through with downward adjustments because of the potential unawareness (due to monitoring difficulties) or

15

unconcernedness of shareholders on subordinates being ‘overcompensated’ due to luck (Merchant and van der Stede, 2012).

2.3 The oil price impact on firms in the O&G industry 2.3.1 O&G industry structure and activities

The O&G industry and its processes and activities are revolved around retrieving petroleum (and gas) from the natural landscape and converting it into a usable end-product for consumers or into an industrial raw material (World Bank, 2009). An overview of the Standard Industrial Classification (hereafter SIC) codes relevant to the O&G industry can be found in Table 1.

Table 1+

An overview of the Standard Industrial Classification (SIC) codes encompassing companies involved in O&G industry activities.

SIC Code Classification Further Explanation

1311 Crude Petroleum & Natural Gas Companies with exploration,

development and production activities related to retrieving crude oil and natural gas up until the point of transportation from the operating site.

1321 Natural Gas Liquids Companies primarily engaged in

producing liquid hydrocarbons from oil and gas field gases

1381 Drilling O&G Wells Drilling wells for O&G field

operations on behalf of others (on a contractual or fee basis)

1382 O&G Field Exploration services Companies engaged in

geophysical, geological and other exploration services for oil and gas on behalf of others (on a contractual or fee basis)

16

2911 Petroleum Refining Companies engaged in producing

fuels (gasoline and kerosene), lubricants and components for petrochemical products.

The industry is generally structured in accordance with the different steps in the O&G value chain, with the main industry sectors labeled as Upstream, Midstream and Downstream (API, 2017). The Upstream sector encompasses exploration, development and retrieval activities relevant to obtain crude oil and natural gas. The Downstream sector includes refining- and other similar activities to get to the end product (e.g. gasoline, kerosene, lubricants and industrial raw materials), as well as marketing, sales and distributions activities to meet customers’ needs. The Midstream sector covers transportation (in its various forms) and storage activities and is a crucial link between Upstream and Downstream activities. This is because this sector is concerned with preserving, marketing and transferring the retrieved crude oil and natural gas from drill locations to processing facilities (World Bank, 2009 ; PSAC, 2017). However, there is no individual SIC code that identifies the Midstream sector. This could be because the activities in this sector are often intertwined with, and reported as Downstream activities (OPEC, 2012). Besides companies specializing in certain activities, there are also companies (labelled Vertically Integrated companies) which are active in all steps of the O&G value chain. Examples include British Petroleum, Exxon Mobil and Royal Dutch Shell.

2.3.2 Oil price and profits in the O&G industry

Previous research has found the oil price to be one of the primary determinants of O&G company performance, profits and market valuation. Bertrand and Mullainathan (2001) point out that oil price fluctuations lead to corresponding large changes in O&G industry profits in their study of the largest companies in the O&G industry in the U.S. between 1977 and 1994. A study by Dayanandan and Donker (2011) come to the same conclusion in their analysis on O&G companies in the U.S. from 1990 till 2008. Moving on, Boyer and Filion (2007) find the oil price to be one of the main risk factors for the O&G industry in Canada in their study on the financial determinants of company stock returns. They argue that the oil price has a pervasive impact on O&G companies as it directly influences revenues, profits, investments and cash flows. Their analysis indicates stock returns to be significantly positively correlated with the oil price level. A study by Elyasiana et al. (2011), analyzing the impact of the oil price on various industries in the

17

U.S., also finds a significant positive correlation between company stock returns of the oil-related (both oil extraction and petroleum refinement sectors) industry and the oil price. The authors point out that this association is the strongest for the oil extraction sector. Elyasiani et al. (2011) also find industries in which oil is an input in the production process, as well as industries that produce a commodity that acts as a substitute for oil, to be responsive to the oil price. Moving on, Ramos and Veiga (2011) provide further support for the positive association between the oil price and O&G industry performance at the global level, by analyzing this phenomenon in 34 countries around the world. However, the authors do note that company share prices are more sensitive to increases in the oil price as opposed to decreases in the oil price.

A deeper understanding of why the oil price and O&G firm valuation is correlated is given by Berry and Wright (2001). A company’s share price and subsequent market capitalization is based on investors’ expectations on the company’s future viability. One primary concern of investors is the company’s ability to generate a consistent growth of profits. Amongst others, investors make a thorough estimate of the potential cash flows and profitability that are likely to come out of future company operations (Berk and Demarzo, 2013). The information gained from this analysis is used to make pricing decisions related to the company’s shares. Berry and Wright (2001), focusing on large (as well as vertically integrated) O&G companies, indicate that a primary component of estimating a company’s future cash flows in the O&G industry relates to the size of the company’s proven and probable O&G reserves (measured in numbers of barrels) and the company’s capability to efficiently discover new oil fields to expand these levels of O&G reserves. The investors’ valuation of these two components is heavily dependent on the oil price level. Regarding O&G reserves, ceteris paribus, a lower oil price level means these reserves are also less valuable. The company’s ability to efficiently expand their O&G reserves is also dependent on the oil price since costs related to exploration and production are relatively fixed, whilst the benefits of these activities decrease when a barrel of oil is worth less. As such, using a cost-benefit analysis, fewer fields are deemed appropriate to retrieve oil from.

Put together, this research supports the notion that the oil price is a primary determinant of O&G industry performance, profitability and market valuation.

2.3.3. Oil price determinants

However, oil is a globally traded commodity. The oil price is driven by the global supply of and demand for oil, with some researchers arguing that derivative market speculations play a role too. Each factor will be discussed in more detail below.

18

Shocks in the supply of oil are caused by various factors. Baumeister and Kilian (2016) and Smith (2009) show that the political instability of major exporting countries plays a role, affecting both the actual supply as well as the expectations of potential disruptions on the future supply of oil. For instance, the supply disruptions from the Iran revolution in the end of the 1970s made the world aware of such supply risk, which led to a buildup of strategic inventories to be able to continue operations in times of major supply disruptions (Smith, 2009). The increase in this so-called precautionary demand caused a sharp oil price increase from about 15 dollars to 40 dollars in April 1980. The Iran-Iraq war that followed later in 1980 led to an actual disruption of the supply of oil from these countries, resulting in a (relatively modest) price increase from 36 dollars in September 1980 to 38 dollars in January 1981. Likewise, the Gulf War caused disruptions in the oil supply from Kuwait and Iraq in the early 1990s, leading to sharp increases in the oil price. However, unlike the price increase during the Iran-Iraq war, Baumeister and Kilian (2016) argue that this supply shock was not the only contributor to the Gulf War price spike. They show that a sharp increase in precautionary demand was equally important. There were concerns that an attack on Saudi Arabia could possibly follow, which thrusted demand for oil inventories in anticipation of a further disruption in the global supply of oil.

The Organization of the Petroleum Exporting Countries (OPEC), an intergovernmental organization of which most of the largest oil exporting countries in the world are members, is another factor that influences the supply side of oil. Their goal is to coordinate the petroleum policies of their member countries to drive the oil price to the level they desire by adjusting their production levels to impact the global supply of oil accordingly (OPEC, 2012). Although the recent introduction of new fracking methods has seen shale oil producers potentially threaten the OPEC dominance in the oil market and its effectiveness at setting the price in the oil market, this extraction method is relatively expensive and only cost efficient when the oil price is high. As such, the OPEC and other major oil exporting countries responded by refusing to adjust their production to this increased supply of oil, with the intention to push the oil price to a level below the cost of fracking oil, thereby shutting down these producers. With oil fracking only recently being seen as a threat to the OPEC dominance, time will tell if the OPEC has lost its ability to set the oil price.

Moving on with the OPEC’s influence on the oil price, Bertrand and Mullainathan (2001) argue that the sharp decrease in the oil price in 1985, as well as the oil price increase in 1979 till 1981, were mainly caused by policy changes of the OPEC and one of its most influential members

19

(Saudi Arabia). The authors therefore find it highly unlikely that the CEO of a single American company is likely to have any influence on the oil price. In support of this, Smith (2009) argues that the conception that the oil market is dominated by a handful of large multi-national corporations is outdated. As of 2007, the eight largest multi-national O&G corporations are responsible for only 12 percent of the global oil production, in comparison to the 89 percent share back in 1969. State-owned companies of major exporting countries are responsible for 50 percent of the output and these state-owned companies hold 70 percent of the world’s proven reserves, in comparison to the 3 percent share of proven reserves these same eight largest O&G companies hold (Smith, 2009).

Although major O&G companies still produce a significant amount of oil, one needs to keep in mind that if e.g. a major company like Exxon Mobil would stop their production altogether, leading to a drop in the supply of oil, the OPEC will respond by increasing their production as an opportunity unfolds to increase revenues. In other words, OPEC members are not running their operations at full capacity, but adjust their production to steer the oil price to the level they desire.

Shocks in the supply of oil, although important, are generally not seen as the most important drivers of the oil price. Numerous studies argue that the main driver of the oil price is the demand for oil (Liu et al., 2016 ; Kilian, 2009 ; Baumeister and Kilian, 2016 ; Juvenal and Petrella, 2014), with some studies specifically pointing to the demand in emerging economies to be the most influential (e.g. Hamilton, 2009). Baumeister and Kilian (2016) argue that most of the oil price fluctuations ranging back to 1973 are largely explained by changes in the demand and precautionary demand for oil. Liu et al. (2016), in their analysis on the determinants of the oil price since 2000, show that the demand for oil from China, but also to a smaller extent the demand from the U.S. (the two largest consumers of crude oil), are the main drivers of the oil price. In their study they find oil supply shocks to be only a minor determinant of the oil price, with derivative market speculations having an even smaller role.

Findings of the impact of derivative market speculations are mixed. In the past few decades the oil market has evolved into a sophisticated financial market with the commodity futures market having a prominent role in the trading of oil (Manera, 2013). Commodity futures’ trading value increased from $3 billion in 2003 to $200 billion in 2008, with oil being the most traded commodity (Liu et al. 2016). Some researchers therefore point to derivative market speculations as being an important determinant of the oil price level since the early 2000s (Morana, 2013 ;

20

Cifarelli and Paladino, 2010). However, some researchers acknowledge that derivative market speculations are secondary to changes in the demand for oil (Juvenal and Petrella, 2014). Other researchers dismiss the claim of the prominent role of speculation. They attribute the oil price to be driven by the strong economic growth and consequent increased demand of emerging countries such as China, with speculation only playing a minor role (Alquist and Gervais, 2013 ; Liu et al. , 2016). Others find no association between derivative market speculation and the oil price altogether (Fattouh, 2012 ; Smith, 2009).

In conclusion, it seems unlikely that even one of the major integrated O&G companies, such as Exxon Mobil, can have any influence on the oil price. With the demand for oil seeming to be the primary determinant of oil prices and with research on the supply side of oil emphasizing the importance of other factors, one can make the conclusion that the oil price is an uncontrollable factor for CEOs of individual U.S. O&G companies. As such, the financial performance measures on which an executive is evaluated are substantially affected by factors outside his control, making them less effective in capturing the executive’s true contribution to the organization.

2.4 Hypotheses

Thus an environment in which financial performance measures (both accounting-based and market-based) are particularly noisy is that of the O&G industry. With the oil price being such a big determinant of the industry’s profitability and market valuation, the O&G industry could be an interesting setting to test whether subjectivity is used to address both the positive and negative influences of external factors on performance measures used in executive’s performance evaluation process and subsequent compensation. The aim of this study is to conduct an analysis on the O&G industry in the U.S. from 2000 till 2015. The chosen timeframe has some interesting oil price trends which could be appropriate for the analysis in this study, as can be seen in Graph 1 below. The increasing trend up until the beginning of the financial crisis (2008), as well as the sharp fall from the summer of 2014 onwards, are periods in which the application of subjectivity in the executive performance evaluation process could be most prevalent.

21

Graph 1: Historical Oil Price Trend

An overview of the historical oil price level during the timeframe 2000 until 2015, retrieved from macrotrends.net (2017). The price level (in U.S. $) reflects the price of one barrel of WTI (West Texas Intermediate) crude oil. A grey column indicates a period of recession.

This study aims to determine whether executive compensation contains more subjective components in the O&G industry in comparison to other industries. After establishing this, the analysis will try to uncover the reasons as to why the use of subjectivity is more prevalent in the O&G industry. Although it seems likely that the oil price is the culprit, to this day (to the best of this author’s knowledge) there is no empirical research that supports such a relationship. This study proposes four hypotheses that will be tested in the analysis.

Following previous research one would suspect that the use of subjectivity would be more common in the O&G industry than in other industries. That is, the O&G industry is an industry in which performance is substantially impacted by an external uncontrollable factor. The consensus is that no single U.S. O&G company can have any influence on the oil price (e.g. Bertrand and Mullainathan, 2001). An external factor is thus creating a lot of noise in executive performance measures when these are based on aggregated accounting- or stock market measures. As such, the performance measures are less reflective of the contribution of executive

22

effort and decisions to company performance and profitability (Merchant and van der Stede, 2012). Executives could get demotivated when the oil price is in a prolonged decreasing trend or could lose proper stimulants to make decisions in alignment with shareholder interests in periods of increasing oil prices. To combat this, subjectivity can be applied to filter out noise and present a more accurate picture of the executive’s performance, thereby increasing the effectiveness of incentive plans in executive compensation packages (Gibbs et al., 2004 ; Bushman et al., 1996). This leads to the following hypothesis:

 H1: The use of subjectivity in executive performance-based compensation is more common in the O&G industry than in other industries.

Moving on, subsequent analyses are conducted to determine why the use of subjectivity is more prevalent in the O&G industry (if indeed the first hypothesis is supported). Previous research on O&G industry performance indicates that the oil price is the primary determinant of industry performance (e.g. Berry and Wright, 2001). Although this hints towards O&G companies’ BoD applying subjectivity in response to noise in executive performance measures attributable to the oil price, prior empirical research has not provided evidence to support this link. One would expect there to be a relation between the oil price and the use of subjectivity in an executive’s performance evaluation. As already indicated, the oil price is an external uncontrollable factor that creates a lot of noise in executive performance measures when these are based on aggregated accounting or stock-market measures. With company profit and market valuation being so heavily impacted by the price level of the core commodity of the industry’s business, it seems likely to expect the use of subjectivity to increase the larger the (annual) oil price change. Graph 2: Annual growth trends of macroeconomic variables

This graph depicts the annual percentage change of the WTI oil price (in US$ per barrel), U.S. GDP (in billions of U.S. $) and the S&P500 Index (index value). Figures were retrieved from Datastream.

23

This relationship could also hold for non-O&G industries, although one would expect it to be the strongest for the O&G industry. As can be seen in Graph 2, the oil price, economic growth and to a lesser extent the stock market growth trends are quite similar. This is no surprise, as oil is an important commodity for many non-O&G industries and generally reflects the broader economic environment (e.g. Chang and Lai, 2013). Although it is not the primary focus of this study, a significant positive association between the oil price and the use of subjectivity in non-O&G industries could be explained following this line of thought. That is, the oil price is a reflection of the economic environment and the use of subjectivity in non-O&G industries is actually driven by this factor rather than the oil price. One also needs to keep in mind that there are certain industries, although smaller in numbers, which actually prosper when the economy (and the oil price) is in a downward trend. Examples are budget retailers and discount stores, as well as companies which provide restoration services.

Overall this leads to two hypotheses, which are presented below.

 H2: The oil price has an influence on the use of subjectivity in executive performance-based compensation.

 H3: The oil price has a larger influence on subjectivity use in executive performance-based compensation in the O&G industry than in non-O&G industries.

Moving on, the general finding of prior research on the (ex-post) use of subjectivity is that it is mainly applied to soften the impact of negative consequences of external uncontrollable factors (Gibbs et al., 2004 ; Bertrand and Mullainathan, 2001 ; Höppe and Moers, 2011 ; Bol and Smith, 2011). That is, upward adjustments to subordinate performance and compensation are made in times of bad company performance to reduce the risk subordinates face when their compensation is based on (noisy) performance measures (Gibbs et al., 2004). However, prior research fails to find evidence on superiors making downward adjustments to subordinate performance and compensation when company performance is good (Merchant and van der Stede, 2012). This so-called asymmetric use of subjectivity is explained to exist because superiors find it problematic to deny rewards to subordinates in times when the company is doing well. Arguments as to why this is include costly confrontations between the superior and the subordinate, fears of a damaged working relationship, and subordinates perceiving these downward adjustments as superiors raising performance targets when performance turns out to be better than expected (Bol, 2008 ; Merchant and van der Stede, 2012). The potential lack of awareness by a company’s owners that employees are being rewarded in excess of their true

24

contribution provides reasoning for superiors to not make downward adjustments to subordinate performance and compensation (Merchant and van der Stede, 2012).

However, the O&G industry could be a setting where the use of subjectivity is not asymmetric. One does not have to be a scholar to understand that the oil price level reflects O&G industry performance. Media reports and analysts’ forecasts all tend to agree that O&G industry profits are high when the oil price is high and vice versa (Forbes, 2016).Thus it seems that company owners are aware of this matter and therefore are more likely to critically evaluate whether executives are rewarded according to their actual contribution to organizational objectives. Therefore, the O&G industry could be an interesting setting to determine whether the BoD applies subjectivity to adjust performance measure outcomes when noise is in favor of the subordinate too. By doing this in addition to when noise has a negative impact, one increases the accuracy of the performance evaluation process regarding the total contribution and the effects of executive decisions and effort on the company. This maintains the effectiveness of the provided incentives, as well as reducing wasteful spending. The final hypothesis related to this matter is presented below.

 H4: The use of subjectivity in O&G executive performance-based compensation does not differ during periods of increasing oil prices and periods of decreasing oil prices.

Although the proposed methodology to investigate the level of subjectivity in executive compensation packages lacks the necessary detail to distinguish which form of subjectivity has been applied, one would expect the higher level of subjectivity in the O&G industry to be attributable to the more frequent ex-post adjustments to compensation. But a higher level of subjectivity use could also be due to the BoD of O&G companies applying subjective weights to ex-ante specified performance measures more often. These adjustments are made to reduce the influence of an uncontrollable factor (the oil price) on the executive’s performance and compensation.

25

3. Methodology

In short, this study consists of archival research using a two-stage regression analysis. The first stage will establish the level of subjectivity applied to executive compensation, after which the second stage will investigate:

a) Whether more subjectivity is used in the O&G industry in comparison to other industries and if this larger use of subjectivity can be explained by changes in the (annual) oil price trend.

b) Whether the level of subjectivity used in the O&G industry differs for periods of increasing oil prices and periods of decreasing oil prices.

3.1 Sample selection

The sample consists of panel data covering companies in various industries in the United States during 2000 until 2015. The starting point of the data collection is to identify and collect compensation, in all its forms, granted to CEOs of listed companies located in the U.S. from the ExecuComp database. Executive compensation is usually reported by companies on an annual basis, thus this study will adopt the same interval. The reason for the cutoff being the year 2015 is that proxy statements and annual reports for the year 2016 have not yet been publicly released by some companies. Information on the variables of interest will therefore not be available in the databases for this year. Reasons for focusing on the U.S. are those of data availability as well as being able to control for country-specific characteristics which may have influence on the use of subjectivity. That is, country-specific regulation regarding executive remuneration or company specifics such as the BoD composition could lead to unnecessary distortions in the analysis. The benefit of focusing on one country is having a more homogenous sample, in which the relationship between executive compensation, subjectivity use and the influence the oil price has on this relationship can be determined more accurately.

Variables retrieved from ExecuComp include the company’s name, company identifiers (GVkey and CUSIP numbers), the company’s SIC code (industry classification), the CEO’s full name, Executive ID number, the ID number for each executive/company combination, Date Became CEO and Date Left as CEO. Variables related to granted CEO compensation that were retrieved include Total Compensation (Salary + Bonus + Other Annual + Restricted Stock Grants + LTIP [Long-term incentive plan] Payouts + All Other + Value of Option Grants), (Discretionary) Bonus and the Non-Equity Incentive Plan Compensation (Formula Bonus). By

26

adding an annual CEO flag as a conditional statement to the database query, only CEO compensation was retrieved. All compensation data of CEOs from banks and other financial services industries (SIC code 6000-6999) were dropped from the sample.

Company specifics will be used in both (regression) stages of the analysis. These specifics will be used to determine what part of the granted CEO compensation can be explained by observable financial measures of performance (Stage 1) and to control for other factors which could explain the level of subjectivity in CEO compensation packages (Stage 2). These company specifics were retrieved from the Compustat database using the GVkey company identifier that came along with the executive compensation data retrieved from ExecuComp. The variables retrieved include Total Assets, Total Liabilities, Net Income, Net Sales, Book Value per Share and Closing (Market) Share Price (end of fiscal year).

Lastly, macroeconomic and other economic fundamentals were retrieved from the Datastream database. These variables are the WTI Crude Oil price (monthly and annual intervals), the U.S. GDP and the S&P500 index (annual intervals).

Table 2: Variable description+

Table 2 presents the variables used in the analysis, with additional information provided on the proxies used for these variables as well as the steps in creating these proxies. The W _{superscript indicates winsorizing was applied to the variable at the 1st and 99th} percentile. The A _{superscript indicates the absolute value of the variable was used in the analysis. All executive compensation} components and company balance sheet statement items (e.g. Total Assets, Net Income) are in millions of U.S. dollars.

Variable Definition Proxy Calculation

Main Analysis

TO_COMPW _{Total Compensation}

ACC_RETW _{Accounting performance ROA (Return on} assets)

Net income/Total Assets

MKT_RETW _{Market performance Stock market return [Share pricet – Share} pricet-1]/Share pricet-1

SUBJECTIVITYA _{Subjective component of} total compensation

Stage 1 residual Absolute value of Stage 1 residual

CEOINFLUENCEW _{CEO influence CEO tenure (in years) Fiscal Year – Date became}