1 Amsterdam Business School
Do equity grants impact earnings manipulation within oil industry downturn
Name: Martijn van Laer Student number: 11156104
Thesis supervisor: Dr. P. Ghazizadeh Date: 26 March 2017
Word count: 10154, 0
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2 Statement of Originality
This document is written by student Martijn Floris van Laer who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Abstract
This thesis explores the effect of CEO equity ownership on the level of earnings manipulation at firms active in the oil industry. The effect of a sudden 48% decline in oil prices in the fourth quarter of 2014 has been taken into consideration when examining the magnitude of earnings manipulation during a period from 2011 to 2016. The observed exogenous oil price shock and its impact on ROA and Book-to-Market ratios explains in part the increase in M-score (Beneish, Lee, & Nichols, 2013). The panel regression applied does however not support my first hypothesis as there is no consistent and significant effect on the M-score in the period post the oil price shock. I do find a significant effect in the interaction between CEO stock ownership and the M-score when comparing the period of low oil prices in 2015 and 2016 with the prior period of relative high oil prices. However, it seems that CEO stock ownership does increase the M-score rather than decreasing it. This finding does not support my second hypothesis, based on Jensen and Meckling (1976) assertions that stock ownership aids the resolution of agency
conflicts. This study adds to the understanding of CEO equity ownership in an environment of an earnings shock and in a period after introduction of tighter regulation (i.e. SOX 2002, Dodd-Frank Act). Furthermore, this research adds to the understanding of the working of the M-score in extreme situations.
4 Contents
1 Introduction ... 6
2 Literature review ... 9
2.1 Agency model ... 9
2.2 Corporate governance and CEO equity grants ... 9
2.3 Corporate governance and CEO duality ... 11
2.4 Earnings manipulation ... 11
2.5 Hypothesis development ... 13
3 Research methodology ... 15
3.1 Type of research ... 15
3.2 Beneish M-score ... 15
3.3 CEO equity compensation ... 15
3.4 Sample data set ... 16
3.5 Variables ... 17
3.5.1 Dependent variables ... 17
3.5.2 Independent variables ... 19
3.5.3 Control variable: CEO duality ... 19
3.5.4 Control variable: ROA ... 20
3.5.5 Control variable: Book-to-Market ... 20
3.6 Research method ... 20
4 Empirical research ... 22
4.1 Descriptive statistics ... 22
5
4.3 Regression analysis ... 26
5 Conclusion ... 29
6 Bibliography ... 31
7 Appendices ... 35
7.1 Appendix 1: NYMEX daily closing Crude Oil price ... 35
7.2 Appendix 2: Results for the Leven’s homogeneity test ... 35
7.3 Appendix 3: Results of the tests for multicollinearity ... 36
7.4 Appendix 4: SIC industry codes at the two-digit level of the sample firms ... 37
7.5 Appendix 5: Firms observed with a negative gross margin ... 38
6 1 Introduction
Economists long argued that a strong linkage between CEO compensation and performance mitigates the agency problem which is caused by separation of ownership and control (Jensen & Meckling, 1976). This view has contributed to a significant increase in total CEO compensation in the past decades whereby stock options and restricted stock grants are the main driver of this growth (Conyon, 2006) (Bushman & Smith, 2001).
Dechow et al (1995) describe that most earnings misrepresentation occurs in an attempt to influence stock price for the reason of avoiding adverse compensation and career consequences. They also find that firms with extreme financial performance are likely to engage in earnings management. In a survey conducted by Dichev et al (2013) this view is confirmed. Of the respondents, 89% say that executive compensation leads to earnings management, and 80% of the respondents believe that senior managers misrepresent earnings to avoid career
consequences.
Accounting scandals such as Enron Corporation, MCI WorldCom seriously damage
confidence and economic performance. In the years 2000-2002, fraudulent financial reports have damaged the U.S. economy, contributing to USD 7 trillion lost by U.S. pension plans (Siebert, 2002). Earnings manipulation is harmful since companies that ‘look’ like past earnings
manipulators will also earn lower future returns (Beneish, Lee, & Nichols, 2013).
Within the tightening legislation such as Sarbanes–Oxley Act of 2002 (SOX, 2002) and Dodd–Frank Wall Street Reform and Consumer Protection Act 2010 (Dodd-Frank, 2010), corporate governance and executive compensation among other topics have been addressed. It was argued and recognized that, with large stock-based bonus at risk, managers were pressured to meet earnings targets. Since the introduction of SOX, stock options are treated as part of the compensation package. This improves the transparency in the monitoring of the total CEO compensation scheme. The Corporate and Criminal Fraud Accountability Title describes specific criminal penalties for manipulation, destruction or alteration of financial records or other
interference with investigations, while providing certain protections for whistle-blowers. Furthermore, the White Collar Crime Penalty Enhancement is aimed to improve the ability to prosecute ‘white-collar crimes and conspiracies’.
7
In the context of tightened regulation, it is interesting to examine the impact of CEO equity compensation in relation to earnings manipulation. This research does not aim to investigate the direct impact of SOX and Dodd-Frank on the magnitude of earnings manipulation but instead investigates, within the setting of more strict regulation, the impact of CEO equity ownership on earnings manipulation. Secondly this study will add to prior research combining CEO option ownership as well as CEO stock ownership. O’Connor (2006) finds that large CEO stock option grants were sometimes associated with a lower incidence of fraudulent reporting and sometimes with a greater incidence of fraudulent reporting, depending upon the frequency of audit
committee meetings and CEO duality. Adding to the research of O’Connor, both CEO option ownership and CEO stock ownership are brought into this research introducing a more comprehensive view on CEO stock compensation and its impact on earnings manipulation.
Globally, oil companies have reduced exploration and production spend by 23% in 2015 (Barclays, 2016). Including impairments, earnings of the top 50 United States oil companies turned into nett losses in 2015 (Byers, 2016) which continued in 2016. This thesis will focus on companies active in the oil sector as it provides a setting of extreme financial performance caused by a 48% drop of crude oil prices in the period 2015 – 2016 compared to the period 2011-2014. This event allows for a panel regression on firms which all suffered from this effect. The central question of research is: “Does CEO equity compensation have an impact on an assumed increased risk of earnings manipulation in the oil industry in the current environment of low oil prices”.
The first focus of the thesis is to investigate the impact of the period pre- and post oil price shock and to investigate its impact on the M-score. The second objective is to examine the impact of CEO equity compensation on earnings manipulation in the period of the structural low oil prices since 2015, compared to the level of earnings manipulation prior to the oil price shock. The M-score methodology of Beneish (1999) will be applied to test the level of earnings
manipulation for the firms in scope of this research.
North American firms, active in the oil sector during the period 2011 to 2016 have been selected. Firms of which financials are incomplete over the years or firms with a market cap of less than USD 50 million and sales and total assets of less than USD 100 million have been excluded from this research. This has resulted in 289 firm years of financials in scope of this research. In the panel regression, I do not find that a period of low oil prices in 2015 and 2016
8
lead to a higher M-score. I do find a significant effect in the interaction between CEO stock ownership and the M-score of firms when comparing the period of low oil prices in 2015 and 2016 compared to its prior period. However, it seems that CEO stock ownership does increase the M-score rather than decreasing it. This finding is not in line with Jensen and Meckling (1976) assertions.
The following chapter reviews the prior literature, examining the relationship between executive incentives and earnings manipulation. The hypotheses formulation is derived from the introduction and literature review and concludes chapter 2. Chapter 3 depicts the data collection process, the variables and methodology used to empirically test the hypotheses. In chapter 4 the descriptive statistics, the correlation matrix and observations are described. A panel regression test has been conducted to test the hypotheses in this chapter. Chapter 6 provides further analysis of the results and will describe the limitations of the research as well as future research
9 2 Literature review
2.1 Agency model
Agency conflicts in firms result from the separation of ownership and control, conflicting objectives and the information asymmetry between owners and managers (Jensen & Meckling, 1976). Because of these agency conflicts, and given that managers have sufficient latitude in applying accepted accounting procedures, executives are likely to have incentives to take actions that maximize their utility, even when those actions do not maximize shareholder wealth (Dey, 2008).
As mentioned in the introduction of this thesis, CEO equity compensation is widely adopted to align the interest of executives and shareholders to reduce the agency problem.
However, the view that the equity component in CEO compensation aids the reduction of agency conflicts is not unambiguous. First, exogenous effects may adversely impact the CEO wealth which may conflict with the idea of rewarding executives for value optimisation. Second,
Bhojrai et al. (2009) find that managers with strong equity-based incentives will cut discretionary expenses and manage accruals to meet and beat the analysts’ forecasts. While such actions result in higher stock market returns in the short term (i.e., in the subsequent year) these actions result in lower longer term returns (i.e., returns over the next three years).
In research conducted by Armstrong et al (2010), it is concluded that various prior studies provide mixed result on empirical tests with regards to the effectiveness of CEO equity
incentives and reduction of accounting irregularities. They find some evidence that accounting irregularities occur less frequently at firms where CEOs have relatively higher levels of equity incentives. Since equity incentives are widely accepted and gaining importance in the CEO compensation package, the view that CEO equity ownership helps reducing the agency problem is followed in this thesis.
2.2 Corporate governance and CEO equity grants
Larcker et al. (2007) describe corporate governance as a set of mechanisms that influence the decisions made by managers when there is a separation of ownership and control. As such Corporate Governance helps to ensure that companies are directed and managed to create value for their owners, while fulfilling responsibilities to other stakeholders.
10
Through the implementation of monitoring mechanisms by shareholders and board of directors or through outcome based incentives, agency conflicts are mitigated. Outcome based incentives in the form of equity is a corporate governance mechanism aimed at reducing friction between outside investors and corporate management control. Jensen and Meckling (1976) describe that equity awards are an important and effective way to align executive interest with those of the shareholder. Bushman & Smith (2001) describe that the total sensitivity of executive wealth to changes in shareholder wealth has become dominated by executives’ stock and stock option portfolios, as opposed to cash compensation or other components of executives’ pay packages.
(Cheng & Warfield, 2005) describe the various forms in which equity is granted to executives being option grants, exercisable stock options, unexercisable stock options, stock grants and stock ownership. When a manager is granted options, these are usually not exercisable until three or four years later. When options become exercisable (i.e., vested), managers can choose to hold exercisable options or choose to exercise the options and hold shares instead. To finance option exercises, managers generally sell the shares received from exercising options. Managers are also awarded restricted stock, which, like option grants, does not vest (i.e., cannot be traded) until after three or four years. Another way to obtain shares is through open market purchases. All cost associated with the granting of options and stock to executives as part of an incentive plan is considered part of CEO remuneration and subject to annual financial disclosure requirements. The different forms of equity ensure that executives wealth is sensitive to firm performance at all times.
Average annual CEO stock ownership market valuation was used to incorporate potential wealth effects and downside risk, which, according to prospect theory (Kahneman & Tversky, 1979), make CEOs more risk-averse (Sanders, 2001). Sanders (2001) noted, for example, that although CEO stock options associated with upside potential promote risk-taking behaviour, CEO stock ownership instead promotes risk averse behaviour. Thus, a higher average annual CEO stock ownership market valuation may be associated with a lower incidence of fraudulent financial reporting. Moreover, increased CEO stock ownership wealth is also associated with higher firm performance (Kay, 2004; McGuire & Matta, 2003).
11
The effectiveness of equity grants depends on corporate government structure (O.Connor, Priem, Coombs, & Gilley, 2006), the mix of equity, in- and out-of-the-money stock options (Zhang, Bartol, Smith, & Pfarrer, 2008) and the industry ‘growth’ environment. Li & Kuo (2017) find a positive correlation between CEO equity compensation and earnings which is more prevalent in an environment of low growth opportunities. Equity grants are an accepted corporate governance mechanism which, in combination with proper oversight should align share holder long term value maximization objectives. In the next paragraph, CEO dualism is introduced as corporate governance indicator which will be part of my research.
2.3 Corporate governance and CEO duality
CEO duality refers to a board leadership structure in which the Chief Executive Officer is the Chairman of the Board. CEO duality has consistently been recognized as a conflict of interest in corporate governance as an indicator of CEO power (Coles & Hesterly, 2000). O’Connor et al (2006) describe that although board members are charged with governing firms, and ensuring high levels of firm performance, they may fail to do so effectively in the presence of board duality, even if they do not formally serve on the committees charged with those responsibilities. This research documents the relation between CEO option ownership, CEO dualism and
Earnings manipulation. O’Connor (2006) research finds that with no CEO stock options, CEO duality is the dominant effect of earnings manipulation in line with arguments that CEO duality indicates CEO power and that some CEOs may use such power to circumvent corporate
governance processes.
2.4 Earnings manipulation
Security analysts, firm managers and investors all devote a great deal of attention to firms reported earnings due to its stewardship role (Chan, Chan, Jegadeesh, & Lakonishok, 2001). Rowchowdhurry (2006) defines real activities manipulation as departures from normal
operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations. In case managers engage in earnings manipulation more extensively than is normal given their economic circumstances, with the objective of meeting/beating an earnings target, they are
12
engaging in real activities manipulation (Zhao, Chen, Zhang, & Davis, 2012) (Murphy &
Zimmerman, 1993). Well known cases of earnings manipulation are Enron Corporation and MCI WorldCom.
Illustrative of earnings challenges in the oil sector, which is subject of research in this thesis, is the reserve matter Shell faced in 2004 when it settled for USD 120 million in relation to a SEC enforcement action for overstating proven hydrocarbon reserves (SEC, 2004). Earnings manipulations are damaging for firms and could lead to bankruptcy and is therefore prosecuted by the SEC Accounting and Auditing Enforcement. Releases suggest a degree of managerial intent that is consistent with a legal claim of fraud defined within the context of SEC Rule 10b-5 or a lesser degree of fraudulent intent defined as misreporting. The Securities Act of 1934 section 10b-5 has been employed for SEC Accounting and Auditing Enforcement proceedings. The Securites Act is enforced through prosecution of executives who are believed to artificially inflate or depress stock prices through earnings manipulation (SEC, 1934). Many plaintiffs in the securities litigation field plead violations of section 10(b) and Rule 10b-5 as a "catch-all" allegation, in addition to violations of the more specific antifraud provisions.
CEOs engaging in earnings manipulation face the risk of prosecution, but at the same time, earnings are supposed to be managed by executives to optimise the firms’ performance or in the case of the oil sector in 2015, to align operating margin to a declining revenue due to the oil price shock. As such, there is a thin line between earnings manipulation and earnings management. Earnings management can in turn be identified as discretionary accrual based earnings management and real activities manipulation (Healy & Wahlen, 1999), (Dechow, Sloan, & Sweeney, 1995), (Dechow, Ge, Larson, & Sloan, 2010), (Dechow, Kothari, & Watts, 1998).
Accrual based earnings management relate to income smoothing by means of accounting intervention. These misstatements have no direct cash flow impact. Accruals can be divided in discretionary and nondiscretionary (Jones, 1991). Real activities manipulation is defined as management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds. After the implementation of SOX in 2002, accrual based earnings management decreased in favour of real earnings management (Cohen, Dey, & Lys, 2008). Both accrual based earnings management as well as real earnings
13
The risk of earnings manipulation, or fraud, prosecuted by the SEC is the focus of this thesis and measured by the M-score as proxy for firms prone to fraud.
2.5 Hypothesis development
As discussed in the previous paragraphs, CEOs engage in earnings manipulation to optimise or preserve personal wealth or to mitigate reputational damage rather than working towards the optimisation of the firms’ value in best interest of the shareholders. CEO equity compensation in a mix of option grants and stock grants with spread vesting periods aims to align the interest between shareholder and CEO (Jensen & Meckling, 1976). From paragraph 2.2, it can be derived that the impact of equity incentives is not unambiguous but for this research it is hypothesised that CEO Equity compensation supports the alignment of interests between principal and agent and reduces the risk of earnings manipulation. In other words, when CEO equity ownership increases, relative to the total remuneration package earned in the year of observation the earnings manipulation index goes down. In the period 2015-2016, with oil prices trading at around 40-50 USD per barrel, earnings in the oil industry are under strain compared to the period 2011-2014 when oil prices ranged around 90.00-100.00 USD per barrel. Before testing the effect of CEO equity ownership and effect on the M-score and as described in the introduction, it is tested whether there is a significant increase in the M-score driven by an extreme outside driven event which potentially drives CEO to manipulate earning (Zhao, Chen, Zhang, & Davis, 2012).
H1: The M-score is significantly higher for the oil sector in a period of low oil prices compared to high oil prices
With the expectation of a noticeable M-score increase (Beneish, 1999) in the period of low earnings performance, it is expected that CEOs with a higher value of equity ownership show significantly less of an increase in M-score compared to CEOs with a lower share of equity ownership, in line with theory (Jensen & Meckling, 1976). Since this research is applied to panel firms pre- and post shock, this expands on research conducted by O’Connor et al (2006), who formed reference portfolios based on various firm characteristics. To investigate this differentiation of M-score with the level of CEO equity compensation a second hypothesis will be tested:
14 H2: The M-score for the oil sector in a period of low oil prices compared to a period with high oil prices is significantly lower when the CEO equity ownership is higher.
Following the research of O’Connor (2006), I will introduce CEO duality in this research as control variable since it is expected that it has a significant effect on the M-score in combination with CEO equity ownership.
The conceptual model of the research conducted is described in the model below (figure 1). The direct relation between stock ownership in the form of the value of awarded equity and option grants and its influence on earnings manipulation is researched.
Figure 1: Conceptual model
ROA, Book-to-Market and CEO duality are added to the model as control variables. Oil price
development
CEO equity ownership
15 3 Research methodology
3.1 Type of research
Annual observations of publicly listed North-American firms active in the oil sector are included in this study. The period 2011-2016 is reviewed since it will capture three years with relative stable oil prices at around USD 100.00 per barrel until 2014 and two observation years of relative low oil prices at around USD 50.00 per barrel. North American companies are selected due to the sample size of industry companies as is captured in CompuStat and ExecuComp. The Beneish M-score model will serve as a model for testing on earnings manipulation risk and a further regression analysis will be conducted to test the relation with CEO equity compensation in line with the hypotheses.
3.2 Beneish M-score
Beneish model is an accounting based earnings manipulation detection model (Beneish, 1999). The calculated M-score provides a model for classifying firms into fraudulent and non-fraudulent reporting firms. Since the publication of the original study the model has attained some notoriety for flagging Enron Corporation well in advance of its eventual collapse. Much of the predictive power of the Beneish model derives from variables that indicate deteriorating fundamentals in fast growing companies (Beneish, Lee, & Nichols, 2013). Earnings
manipulation as determined by Beneish M-score comprise the dependent variable in this research.
3.3 CEO equity compensation
From the ExecuComp database the independent variables are retrieved to establish the fair value of equity ownership relative to the full CEO compensation package. Elements of CEO
compensation investigated are total stock ownership relative to full remuneration package. Total stock ownership is derived from restricted stock grants, in-the-money stock options and out-of-the-money stock options. Total remuneration includes all cash elements and entitlements within the year of observation, performance plans which include valued stock options using the Black-Scholes model at time of issuance, and restricted stock valued on the date of grant. Stock options in ExecuComp are priced using Black and Scholes methodology. Stock options will vest over
16
time as part of a remuneration plan and are not exchange traded. Although this might affect the application of the Black & Scholes model, the pricing as displayed in ExecuComp is applied to value stock option grants in this research.
3.4 Sample data set
The investigation has been carried out by dividing the sample period into two time periods: the period of consistently high oil prices from 2011 to 2014 and the subsequent period of low oil prices from 2015 to 2016. The Sample data is collected from the Wharton Research Data Services (WRDS). Financial statement data for U.S. listed firms (CompuStat) have been
retrieved from the ‘Fundamentals Annual’ database with a research period of 2011-2016. The Sic codes 1311, 1381, 1382, 1389 have been selected to represent the companies active or related to the Oil & Gas industry. These Sic codes represent types of activities included such as
exploration, drilling, oil and gas well operation and maintenance, the operation of natural gasoline and cycle plants, and the gasification, liquefaction, and pyrolysis of coal at the mine site. This major group also includes such basic activities as emulsion breaking and desilting of crude petroleum in the preparation of oil and gas customarily done at the field site. In addition, Sic code 2911 has been selected to cover establishments primarily engaged in petroleum refining and in the production of lubricating oils and greases. Firms with total assets as well as sales of less than $100 million and firms with a market capitalisation of less than $50 million are excluded. This has resulted in 1,333 observations excluded from the research.
From CompuStat’s Executive Compensation Database (ExecuComp), CEO stock holding and CEO option holdings are added to the dataset retrieved from the CompuStat database. This dataset will be used to evaluate the value of the CEO equity ownership at year end, relative to total remuneration of the year. To test on the CEO duality control variable, the Institutional Shareholder Services (ISS) Directors database has been used merged into the dataset of CompuStat.
17 Table 1: Firm sample selection
# Observations (N) # Firms
Data retrieved from CompuStat 2,224 580
Data merged from ExecuComp 433 79
8 1
Data merged from Institutional Shareholder Services 299 60
6 2
299 60
Remove firms with missing observation years -10 -4
Final sample 289 56
Table 1 shows that after merging the sample data set of 580 firms retrieved from CompuStat with ExecuComp and ISS, 56 firms provide a match across the various years. Both ExecuComp and ISS match a limited number of firms with CompuStat resulting in a loss of 524 firms from the CompuStat data retrieval. Nevertheless, 289 firm year observations can be used for testing the hypotheses.
3.5 Variables
3.5.1 Dependent variables
To measure earnings manipulation as dependent variable will be derived from CompuStat data and calculated using the Beneish methodology. Beneish M-score is measured through:
𝑀 = −4.84 + 0.920 ∗ 𝐷𝑆𝑅𝐼 + 0.528 ∗ 𝐺𝑀𝐼 + 0.404 ∗ 𝐴𝑄𝐼 + 0.892 ∗ 𝑆𝐺𝐼 + 0.115 ∗ 𝐷𝐸𝑃𝐼 − 0.172 ∗ 𝑆𝐺𝐴𝐼 + 4.679 ∗ 𝐴𝐶𝐶𝑅 − 0.327 ∗ 𝐿𝐸𝑉𝐼
DSRI (Days Sales in Receivables Index) captures distortions in receivables that can result from revenue inflation. The variable is calculated as follows:
𝐷𝑆𝑅𝐼 =(𝑛𝑒𝑡𝐴𝑅B 𝑅𝐸𝐶𝑇1 𝑆𝑎𝑙𝑒𝑠B 𝑆𝐴𝐿𝐸 𝑛𝑒𝑡𝐴𝑅BGH 𝑆𝑎𝑙𝑒𝑠BGH
18
GMI (Gross Margin Index) deteriorating margins predispose companies to manipulate earnings. According to Beneish (1999) the GMI will be greater than 1 in case the gross margin has deteriorated versus 1-t.
𝐺𝑀𝐼 = 𝑆𝑎𝑙𝑒𝑠BGH− 𝐶𝑜𝑠𝑡𝑜𝑓𝐺𝑜𝑜𝑑𝑠𝑆𝑜𝑙𝑑BGH 𝐶𝑂𝐺𝑆
𝑆𝑎𝑙𝑒𝑠BGH
𝑆𝑎𝑙𝑒𝑠B− 𝐶𝑜𝑠𝑡𝑜𝑓𝐺𝑜𝑜𝑑𝑠𝑆𝑜𝑙𝑑B 𝑆𝑎𝑙𝑒𝑠B
AQI (Asset Quality Index) captures distortions in other assets that can result from excessive expenditure capitalization.
𝐴𝑄𝐼 = 1 −𝐶𝑢𝑟𝑟𝑒𝑛𝑡𝐴𝑠𝑠𝑒𝑡𝑠B[𝐴𝐶𝑇] + 𝑛𝑒𝑡𝑃𝑃𝐸B [𝑃𝑃𝐸𝑁𝑇]
𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠B[𝐴𝑇] 1 −
𝐶𝑢𝑟𝑟𝑒𝑛𝑡𝐴𝑠𝑠𝑒𝑡𝑠BGH+ 𝑛𝑒𝑡𝑃𝑃𝐸BGH 𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠BGH
SGI (Sales Growth Index) manages the perception of continuing growth and capital needs predisposes growth companies to manipulate sales and earnings.
𝑆𝐺𝐼 = 𝑆𝑎𝑙𝑒𝑠B 𝑆𝑎𝑙𝑒𝑠BGH
DEPI (depreciation Index) captures declining depreciation rates as a form of earnings manipulation.
𝐷𝐸𝑃𝐼 = 𝐷𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛𝐸𝑥𝑝𝑒𝑛𝑐𝑒BGH[𝐷𝑃 − 𝐴𝑀] 𝐷𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛𝐸𝑥𝑝𝑒𝑛𝑠𝑒BGH+ 𝑛𝑒𝑡𝑃𝑃𝐸BGH[𝑃𝑃𝐸𝑁𝑇]
𝐷𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛𝐸𝑥𝑝𝑒𝑛𝑠𝑒B 𝐷𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛𝐸𝑥𝑝𝑒𝑛𝑠𝑒B+ 𝑛𝑒𝑡𝑃𝑃𝐸B SGAI (Sales and General Administration Expenses Index) decreasing administrative and
marketing efficiency (larger fixed SGA expenses) predisposes companies to manipulate earnings.
𝑆𝐺𝐴𝐼 = 𝑆𝐺𝐴𝑒𝑥𝑝𝑒𝑛𝑠𝑒B[𝑋𝑆𝐺𝐴] 𝑆𝑎𝑙𝑒𝑠B
𝑆𝐺𝐴𝑒𝑥𝑝𝑒𝑛𝑠𝑒BGH 𝑆𝑎𝑙𝑒𝑠BGH
ACCR (accruals to total assets) captures where accounting profits are not supported by cash profits. Beneish (1999) approached the calculation of accruals from balance sheet analysis since Cash Flow from Operations was not reported at the time. Since this information is now available and deemed to be more consistent compared to the use of balance sheet accruals (Hribar & Collins, 2002), the use of the statement of cash flows rather than the balance sheet is used in line with Beneish et al (2013). Income before Exceptional Items (IBEI) minus Cashflow from
Operations is divided b Total Assets
𝐴𝐶𝐶𝑅𝐼 = 𝐼𝐵𝐸𝐼B[𝐼𝐵] − 𝐶𝐹𝑂B[𝑂𝐴𝑁𝐶𝐹]2 𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠B
19
LEVI (Leverage Index) flags increasing leverage which tightens debt constraints and predisposes companies to manipulate earnings.
𝐿𝐸𝑉𝐼 = 𝐿𝑇𝐷B 𝐿𝐶𝑇 + 𝐶𝑢𝑟𝑟𝑒𝑛𝑡𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠B[𝐷𝐿𝑇𝑇] 𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠B
𝐿𝑇𝐷BGH+ 𝐶𝑢𝑟𝑒𝑛𝑡𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠BGH 𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠BGH
When the manipulation index exceeds -1,78 the relevant firm is suspected to manipulate earnings (Beneish, 1999). The M-score is calculated per fiscal year with comparison to the prior fiscal year. More specifically, the model captures variables that capture financial statement distortions (DSRI, AQI, DEPI, Accruals) or a predisposition to engage in earnings manipulation owing to economic conditions (GMI, SGI, SGAI, LEVI).
3.5.2 Independent variables
The independent variables retrieved from the ExecuComp Annual Compensation database represent the total value of awarded stock and options as reported under FASB 123R. The relative value of the total value of stock options and equity awards have proportioned to the total compensation as filled in with the SEC (Form 10-K) on an annual basis.
𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑣𝑎𝑙𝑢𝑒 𝑠𝑡𝑜𝑐𝑘 𝑎𝑤𝑎𝑟𝑑𝑠 = 𝑆𝑡𝑜𝑐𝑘 𝑣𝑎𝑙𝑢𝑒B[𝑆𝑡𝑜𝑐𝑘_𝐴𝑤𝑎𝑟𝑑𝑠]3 𝑇𝑜𝑡𝑎𝑙 𝐴𝑛𝑛𝑢𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛B [𝑇𝑜𝑡𝑎𝑙 𝑆𝐸𝐶] 𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑝𝑡𝑖𝑜𝑛 𝑎𝑤𝑎𝑟𝑑𝑠 = 𝑂𝑝𝑡𝑖𝑜𝑛 𝑣𝑎𝑙𝑢𝑒B[𝑂𝑝𝑡𝑖𝑜𝑛_𝐴𝑤𝑎𝑟𝑑𝑠] 𝑇𝑜𝑡𝑎𝑙 𝐴𝑛𝑛𝑢𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛B 𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑡𝑜𝑡𝑎𝑙 𝑒𝑞𝑢𝑖𝑡𝑦 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 =𝑂𝑝𝑡𝑖𝑜𝑛 𝑣𝑎𝑙𝑢𝑒 B+ 𝑆𝑡𝑜𝑐𝑘 𝑣𝑎𝑙𝑢𝑒B 𝑇𝑜𝑡𝑎𝑙 𝐴𝑛𝑛𝑢𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛B
3.5.3 Control variable: CEO duality
The Director database from the Institutional Shareholder Services has been consulted whereby on an executive names basis it is determined per year whether the director is CEO and Chairman.
𝐶𝐸𝑂 𝐷𝑢𝑎𝑙𝑖𝑡𝑦 = 𝐸𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡 𝑡𝑖𝑡𝑙𝑒 − 𝐶𝐸𝑂 = [𝐸𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡 𝑡𝑖𝑡𝑙𝑒 − 𝐶ℎ𝑎𝑖𝑟𝑚𝑎𝑛]
20
As mentioned before, CEO duality is expected to show a positive correlation with earnings manipulation.
3.5.4 Control variable: ROA
To mitigate the omitted-variables problem, the ROA and Book-to-Market variables are included in the regression. In the context of earnings management, ROA will be used as a control variable for measuring firm performance and discretionary accruals. ROA is added to the Beneish model and is also applied as control variable by Thiruvadi & Huang (2011).
𝑅𝑂𝐴B =
𝐸𝐵𝐼𝑇B[𝑁𝐼] 𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠B[𝐴𝑇]
3.5.5 Control variable: Book-to-Market
The book to market ratio as control variable is calculated by dividing the book value of equity at the end of the respective firms’ fiscal year divided by the market capitalisation. Market
capitalisation is calculated by multiplying the number of common shares outstanding with the share price at close of the relevant fiscal year. A high Book-to-Market ratio indicates low firm growth opportunities (Kothari, Leone, & Wasley, 2005).
𝐵𝑀B = 𝐸𝑞𝑢𝑖𝑡𝑦𝐵𝑜𝑜𝑘𝑣𝑎𝑙𝑢𝑒B[𝑆𝐸𝑄]
𝐸𝑞𝑢𝑖𝑡𝑦𝑚𝑎𝑟𝑘𝑒𝑡𝑣𝑎𝑙𝑢𝑒B[𝐶𝑆𝐻𝑂 + 𝑃𝑅𝐶𝐶] 3.6 Research method
A panel regression has been conducted to test the linear relationship between the CEO equity holding and the M-score over the course of several periods. Panel data contains multiple observations whereby consecutive fiscal year observations for the same firms are tested. There are four models tested for the linear relationship between dependent and independent variables. As displayed in table 2, the first model displays a regression on the control variables ROA, Book-to-Market and CEO duality. The second model adds the Period to the control variables to test the period effect separating, post oil price shock (2015, 2016) from the prior period (2011 to 2014). In the third model the CEO stock ownership and CEO option ownership are added to the second panel regression model. The fourth model includes the interaction between the dependent variables, CEO option ownership, CEO stock ownership with Period and builds on the third model.
21 Table 2: Regression models
Model 1: Control 𝑀𝑠𝑐𝑜𝑟𝑒 = 𝛼 + 𝛽 ∗ 𝑅𝑂𝐴 + 𝛽 ∗ 𝐵𝑇𝑀 + 𝛽 ∗ 𝐷𝑈𝐴𝐿 + ℰ Model 2: Main 1 𝑀𝑠𝑐𝑜𝑟𝑒 = 𝛼 + 𝛽 ∗ 𝑅𝑂𝐴 + 𝛽 ∗ 𝐵𝑇𝑀 + 𝛽 ∗ DUAL + 𝛽 ∗ 𝑃𝐸𝑅𝐼𝑂𝐷
+ ℇ
Model 3: Main 2 𝑀𝑠𝑐𝑜𝑟𝑒 = 𝛼 + 𝛽 ∗ 𝑅𝑂𝐴 + 𝛽 ∗ 𝐵𝑇𝑀 + 𝛽 ∗ DUAL + 𝛽 ∗ 𝑃𝐸𝑅𝐼𝑂𝐷 + 𝛽 ∗ 𝑂𝑃𝑇𝐼𝑂𝑁 + 𝛽 ∗ 𝑂𝑃𝑇𝐼𝑂𝑁 + ℇ
Model 4: Interaction 𝑀𝑠𝑐𝑜𝑟𝑒 = 𝛼 + 𝛽 ∗ 𝑅𝑂𝐴 + 𝛽 ∗ 𝐵𝑇𝑀 + 𝛽 ∗ DUAL + 𝛽 ∗ 𝑃𝐸𝑅𝐼𝑂𝐷 + 𝛽 ∗ 𝑂𝑃𝑇𝐼𝑂𝑁 + 𝛽 ∗ 𝑂𝑃𝑇𝐼𝑂𝑁
+ 𝛽 𝑂𝑃𝑇𝐼𝑂𝑁 ∗ 𝑃𝐸𝑅𝐼𝑂𝐷 + 𝛽 𝑆𝑇𝑂𝐶𝐾 ∗ 𝑃𝐸𝑅𝐼𝑂𝐷 + ℇ
Before regression analysis is conducted the continuous variables have been Windsorised. The Windsorisation is conducted on 99%, for all continuous variables except AQI and GMI, meaning that the extreme outliers which could influence descriptive statistics and regression are adjusted. As such, the first and last percentile continues variables are changed to minimum and maximum values.
Observations in the Asset Quality Index (AQI), which captures the capital distortions, do result in some cases in extreme values. This occurs when ‘Property Plant Equipment’ and
‘Current Assets’ added together is equal to the Total Assets in either the current or previous year observed. Since these observations adversely impact this research the AQI index is Windsorised at 95%.
The Gross Margin Index (GMI) has been adjusted for extreme values and is Windsorised at 95% as well. Various firms (Appendix 5, table 10) show negative gross margins in various years of observation. Appendix 6, shows the impact of such negative gross margins in a calculation example (Figure 3). Following this example and to avoid misrepresentation a minimum gross margin of 10% has been applied to observations with a negative gross margin.
22 4 Empirical research
4.1 Descriptive statistics
Table 3 provides an overview of annual average M-scores, CEO equity ownership, CEO duality, ROA and Book-to-Market ratios of the firms in scope. The M-score increases noticeably in 2015 compared to 2014. The M-score is used to assess year-on-year movements. In the year 2015, when the oil price shock has impact on the first full year, the average M-score for firms active in the oil industry increases from -2,91 in 2014 to 3,54 in 2015. An M-score higher than -1.78 flags companies prone to fraud according to Beneish (1999). I observe 66 flagged firms in 2015, compared to 7 in the years 2013 and 2014. Movement in accruals (ACCRI) and an
increased gross margin index (GMI) are the main contributors to the M-score increase. An increased GMI signals a decreased margin. It is interesting to find that in 2016 the M-score is -2.84 which is an improvement compared to 2015, however it earmarks 22 firms as prone to fraud suggesting a higher variance, or more exceptional observations of firms having an M-score of -1.78 or higher.
Where oil companies achieved a positive ROA in the years before 2015, this has turned to -19.9% and -9.6% in 2014 and 2015 which indicates a loss position for the combined firms active in the oil industry. Oil prices sharply declined in the last quarter of 2014 and settled in 2015 and 2016 on an average of $48,88 and $43,41 respectively, compared to an average annual level ranging from $94,15 and $98,11 per barrel in the period 2011 to 2013. The Book-to-Market ratio increases in 2014 and 2015 to 1.07 and 1.29 indicating that the Net Equity value at fiscal year-end is above the market capitalisation. In the years 2011-2013 the reported Book-to-Market ratio is between 0.66 and 0.76.
CEO total equity ownership has increased in the years 2014 to 2016 compared to the prior years. The relative share of stock options has declined in favour of equity over the years. Overall CEO equity holding has increased from 50.7% in 2011 to 59.8% in 2016 relative to overall annual compensation. CEO duality, in which case the CEO is also chairman of the board, ranges in the years between 40.4% to 58.5%.
When the Oil price and Book-to-Market ratios deteriorated in 2014, the firms ROA and M-score show an observable negative movement in 2015. From table 3 it can be derived that,
23
with 2014 as a transitional year, the firms active in the oil sector, suffered losses and challenges to balance operational cost in the years 2015 and 2016 compared to the period 2011 to 2014 when oil prices were almost double the rates (Appendix 1).
Table 3: Average M-score, CEO equity, CEO duality and control variables per year
Year 2011 2012 2013 2014 2015 2016
12M avg crude oil$, Nymex 95.11 94.15 98.05 92.82 48.88 43.41
# FIRMS COMPUSTAT 208 213 223 234 224 196 a. Intercept -4,84 -4,84 -4,84 -4,84 -4,84 -4,84 b. DSRI * 0.920 0,77 0,84 0,93 0,82 0,94 1,23 c. GMI * 0.528 0,43 0,50 0,50 0,56 8,31 0,68 d. AQI * 0.404 0,32 0,36 0,36 0,40 0,40 0,40 e. SGI * 0.892 1,01 0,95 0,98 1,08 0,64 0,71 f. DEPI * 0.115 0,07 0,08 0,08 0,08 0,10 0,13 g. SGAI * 0.172 -0,13 -0,17 -0,17 -0,15 -0,24 -0,20 h. ACCRI * 4.679 -0,37 -0,46 -0,46 -0,52 -1,39 -0,62 i. LEVI * 0.327 -0,30 -0,32 -0,33 -0,35 -0,37 -0,32
M-SCORE (sum a-i) -3,04 -3,06 -2,94 -2,91 3,54 -2,84
# ‘Fraud prone’ Firms4 11 7 7 19 66 22
# FIRMS EXECUCOMP 64 66 66 69 69 62
CEO Options ownership 13,9% 11,6% 10,2% 7,1% 7,2% 8,4%
CEO Stock ownership 36,9% 42,4% 44,9% 49,4% 57,9% 51,5%
CEO Equity ownership 50,7% 54,0% 55,1% 56,4% 65,1% 59,8%
# FIRMS ISS 41 47 48 52 51 50
CEO duality % Firms 58,5% 40,4% 52,1% 40,4% 41,2% 46,0%
Return on Assets 6,3% 4,4% 4,7% 4,0% -19,9% -9,6%
Book-to-Market 0,68 0,76 0,66 1,07 1,29 0,86
4 A Firm with an M-score larger than -1,78 is flagged as fraud prone firm from the CompuStat database firm selection.
24
Tabel 4 is presented in three periods and shows the means and standard deviation for the variables. Period 0 represents firms’ reporting in fiscal years 2011-2013, period 1 covers firms reporting in fiscal year 2014 and the years 2015 and 2016 are represented in period 2. With exception of CEO duality with a relative low F-Score there is a significant difference in means between the different periods for all variables which are part of this research.
Table 4: Descriptive statistics
Period 2011 – 2013 (0) 2014 (1) 2015-2016 (2) M SD M SD M SD F p Oil prices 95.78 7.33 92.82 13.49 46.14 7.32 5539.19 0,0000 M-score -3,02 0,89 -2.91 0.84 0,56 9,58 59,15 0.0000 Stock ownership 0.41 0.23 0.50 0.23 0.55 0.19 15.04 0.0000 Option ownership 0.12 0.15 0.07 0.11 0.08 0.11 9.34 0.0001 CEO duality 0.50 0.50 0.40 0.50 0.44 0.50 0.88 0.4139 Book-to-Market 0.70 0.38 1.07 0.76 1.09 1.87 16.74 0.0000 Return on Assets 0.43 0.07 0.02 0.08 -0.13 0.26 147.30 0.0000
When applying Levene’s test, not all variables seem to be homogeneous (Appendix 2). In the regression analysis, the period 2011 to 2014 is compared with the period 2015-2016, the years with the full year impact of the relatively low oil price.
Low crude oil prices and negative returns on assets may drive CEOs to manipulate
earnings. On the other hand, factors such as increasing CEO equity ownership and alterations in the percentage of CEO duality could have an impact on the development of the M-score. In the following paragraphs 4.3 and 4.4 a correlation and regression analysis have been conducted to further investigate the relation between variables.
25
4.2 Correlations
Table 5 shows the Pearsons correlation matrix including significance levels between the regression variables M-score, CEO equity ownership and the control variables CEO duality, ROA and Book-to-Market. It can be deducted from the correlation table that when the M-score increases the overall value of CEO equity ownership increases. Although the total CEO equity ownership is weakly positively correlated with the M-score, the mix between CEO stock ownership and CEO option ownership changes. A moderate negative correlation indicates that increased CEO stock holdings correlates with declining value of CEO option ownership. Since a relative large portion of the total CEO equity ownership is comprised of stock ownership, logically, a strong correlation is observed.
CEO duality weakly correlates to CEO equity ownership. This positive relationship implies that when the percentage of CEOs being Chairman increases and total equity ownership increases. ROA shows a moderate negative correlation with the M-score and a weak negative correlation with CEO equity holdings. A weak negative correlation exists between Market ratio and CEO dualism. As expected a positive correlation exists between Book-to-Market and ROA.
Table 5: Correlations Variable 1 2 3 4 5 6 1. M-score 2. Option ownership -0.1191* 3. Stock ownership 0.2017*** -0.5180*** 4. Equity ownership 0.1538*** 0.0786 0.8121*** 5. CEO duality -0.0230 -0.0087 0.1011 0.1182* 6. ROA -0.5267*** 0.1269* -0.2555*** -0.2112*** 0.0933 7. Book-to-Market 0.0148 -0.0819 0.0961 0.0561 -0.1260* 0.1271*** *p<0.05, **p<0.01, ***p<0.001
26
4.3 Regression analysis
The hypotheses are tested using the panel regression model. A panel regression is suitable to test the development of the M-score for different firms over multiple fiscal years ending in the period from 2011 until 2016. Support is sought for the relationship between the main effect variables, CEO stock ownership and CEO option ownership on the development of the M-score in the period of low oil prices in 2015 to 2016 compared to the prior period. Tabel 5 displays the four models that have been tested. These models assess whether there is a negative relation between CEO equity ownership and the M-score. The effect of the Oil price shock in the last quarter of 2014 affecting 2015 and 2016 compared to the prior period is included in the various models.
Model 1 of the regression represents the linear model testing the control variables ROA, Book-to-Market and CEO duality. Although model 1 does not test the hypotheses, it does help understanding the impact of the independent variables on the M-score. At the same time, it mitigates the omission of variables effect in the interpretation of the regression test. From the F-test the relation between the control variables and the M-score can be derived. The F-F-test (3,230) = 65.51 and significant with p<0.001. Both ROA and Book-to-Market contribute on a
significance level of p<0.001 to the 46.1% of the variability of the M-score. The ROA coefficient of -26.01 is in line with expectation of a negative relation between the M-score and ROA
implying that declining profitability increases the risk of earnings manipulation. The opposite is true for the Book-to-Market coefficient of 2.81 showing a positive relation between the M-score and a declining market capitalisation compared to total equity of the firm.
Model 2 introduces the timing effect of the M-score which is based on the development of the Oil price. This categorical dummy variable flags the observation years 2015 and 2016 in which the oil price is consistently low compared to the prior period of firm observations. The introduction of the period variable shows no significance in relation with the M-score. The addition of the period indicator only adds 0.6% to the R square of 46.7% variability of the M-score. Model 2 tests Hypothesis 1, rejecting the view that in the years 2015, 2016 the M-score is significantly higher, compared to the prior period of observation. The control variables explain most of the movement of the M-score.
In the third model, both the option ownership value and stock ownership value, relative to total earnings of the year, is added. These independent variables test the relation between CEO stock ownership in the two distinct groups of options and shares and its impact on the M-score.
27
Although R square improves marginally with 0.4% to 47.1% compared to the model 2, this cannot be attributed to stock ownership since the standard error is too large.
In the panel regression of model 4 the interaction variables are introduced in addition to the main effect variables. The F-test (8, 225) is 26.85 with a value of p<0.001. The model explains 48.8% of the variance in M-score. With introduction of the interaction variables ‘Stock * Period’ and ‘Option * Period’, Period has a significant (p<0.05) effect on the M-score.
The interaction between CEO stock ownership and Period has a significant (p<0.05) effect on the M-score. Hypothesis 2 seeks support for the assertions made by Jensen and
Meckling (1976) that CEO equity ownership aids the reduction of the agency conflict. As such, I hypothesised that in the period of low oil prices the M-score increases less for firms where CEOs own more shares, relative to their total annual salary. Although we find evidence for a relation between CEO stock ownership, there seems to be a negative relation between stock ownership and manipulation. Hypothesis 2 is therefore rejected.
28 Table 6: Regression estimates
Results from the panel regression analysis with M-score as dependent variable
(1) (2) (3) (4)
VARIABLES Model 1 Model 2 Model 3 Model 4
Control variables ROA -26.01*** -28.48*** -29.02*** -27.22*** (1.985) (2.482) (2.523) (2.574) Book-to-Market 2.81*** 3.04*** 3.14*** 2.96*** (0.505) (0.521) (0.530) (0.527) CEO duality 1.51 1.51 1.50 1.51 (1.046) (1.042) (1.044) (1.031)
Main effect variables
CEO stock ownership -3.14 -4.69
(2.815) (2.877)
CEO option ownership -0.42 1.13
(5.175) (5.347) Period: post 2014 -1.42 -1.21 -7.12** (0.862) (0.879) (3.431) Interaction variables Stock_x_Period 11.48** (5.347) Option_x_Period 1.62 (8.186) Constant -4.83*** -4.58*** -3.20* -2.64 Observations 289 289 289 289 R-squared 0.461 0.467 0.471 0.488 F 65.51 50.18 33.64 26.85 Number of id 56 56 56 56
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
29 5 Conclusion
This thesis aims to answer whether CEO equity compensation has an impact on an assumed increased risk of earnings manipulation in the oil industry in the current environment of low oil prices. The M-score methodology has been applied, inspired by the achievements of Beneish in successful identifying firms prone to fraud.
Although research provides addition and nuance to the view of Jensen and Meckling (1976) in which is asserted that CEO equity ownership aligns shareholders and management objectives, their view is followed in this thesis. To this extent, CEO equity ownership relative to total remuneration has continued to grow since the 1990’s and especially in firms where the equity ownership at CEO level is relatively high, I hypothesise that those firms are less prone to fraud. O’Connor et al (2006) add in their research CEO duality as a moderating factor to the impact of option ownership. The assertion that CEO duality adversely impacts the agency conflict is included in the regression analysis as control variable, next to ROA and Book-to-Market. The setting in which my research is conducted is in a tightened accounting environment, post SOX and post implementation of Dodd-Frank Act and, as such, covers an observation period from 2011 to 2016. In combination with the earnings shock suffered by the oil sector during this period, this allowed for panel regression analysis.
Hypothesis 1, testing the view that an earnings shock makes firms more prone to earnings manipulation, expressed in an increased M-score for the period after the oil price shock is not supported for the firms in scope of my research.
Although the relative percentage of equity compensation has increased on average from the period 2011 to 2016, no supporting evidence has been found that higher equity compensation for firms compared to firms with less CEO equity ownership leads to less manipulation in times of oil price distress. Hypothesis 2 is therefore rejected. In fact, it seems that firms with relative higher share of CEO equity ownership compared to firms with a lower share of stock ownership have a higher risk of fraud. This is contradicting the view of Jensen and Meckling (1976).
This research has various limitations. The Beneish M-score does not provide for the extreme situation of negative gross margin in the calculation of GMI (appendix 6). In future research, the conditions and boundaries in which the M-score has predictive value could be further
investigated. Furthermore, the total size of the sample set, completed with information from the databases of ISS and ExecuComp is limited and has resulted in the selection of relative large
30
firms in this research. From prior research as well as in this thesis, it appears that CEO stock ownership and its impact on aligning CEO and shareholders interest is not unambiguous. Various factors seem to impact the effectiveness of CEO equity incentives. As equity incentives gain importance since the 1990’s, the effectiveness of such programs and interaction with governance structures should be further investigated.
31 6 Bibliography
Armstrong, C., Alan, J., & Larcker, D. (2010, May). Chief Officer Equity incentives and Accounting Iregularities. Journal of Accounting Research, 48(No. 2), pp. 225 - 271.
Arthur, L. (1998). The "Number Game". Securities and Exchange Commission.
Barclays. (2016, 14 January). North American Oil Industry Braces for More Cuts in 2016. David Wethe, 1.
Beneish, M. (1999, Sep -Oct.). The detection of Earnings Maniplation. CFA institute, Vol. 55 (No. 5), pp. 24-36.
Beneish, M., Lee, C., & Nichols, C. (2013, March/April). Earnings Manipulation and Expected Returns. Financial Analyst Journal, Vol. 69(2), pp. 57-82.
Bhojraj, A., Hribar, P., Picconi, M., & McInnis, J. (2009). Making Sense of Cents: An Examination of Firms That Marginally Miss or Beat Analyst Forecasts. The Journal of Finance, pp. 2361–2388. Bushman, R., & Smith, A. (2001). Financial accounting information and corporate governance. Journal
of accounting & economics(32), pp. 237-333.
Byers, D. (2016). US oil and gas reserves study 2016. www.ey.com/oilandgas: Ernst & Young.
Chan, K., Chan, L., Jegadeesh, N., & Lakonishok, J. (2001). Earnings Quality and Stock Return.
National Bureau of Economic Research.
Cheng, Q., & Warfield, t. (2005). Equity Incentives and Earnings Management. American Accounting
Association, 80(No. 2), pp. 441 - 476.
Cohen, D., & Zarowin, P. (2010). Accrual-based and real earnings management activities around seasoned equity offerings. Journal of accounting and economics, pp. 2-19.
Cohen, D., Dey, A., & Lys, T. (2008). Real and Accrual-Based Earnings Management in the Pre- and Post-Sarbanes-Oxley Periods. The Accounting Review, pp. 757-787.
Coles, J., & Hesterly, W. (2000). Independence of the Chairman and board composition: Firm choices and shareholder value. Journal of Management(26), pp. 195 - 214.
Coles, J., Daniel, N., & Naveen, L. (2006). Managerial incentives and risk-taking. Journal of Financial
32
Conyon. (2006, February). Executive Compensation and Incentives. Academy of Management Perspectives,
20(No. 1), pp. 25 - 44.
Dechow, P., & Sloan, R. (1991). Executive incentives and the horizon problem. Journal of Accounting
and Economics, pp. 51-89.
Dechow, P., Ge, W., Larson, C., & Sloan, R. (2010). Predicting Material Accounting Misstatement.
Contemporary Accounting esearch, Vol. 28(No. 1), pp. 17-82.
Dechow, P., Kothari, S., & Watts, R. (1998). The relation between earnings and cash flows. Journal of
Accounting and Economics, Vol. 25, pp. 133-168.
Dechow, P., Sloan, R., & Sweeney, A. (1995, April). Detecting Earnings Management. American
Accounting Association, Vol. 70(No. 2), pp. 193-225.
Dey, A. (2008, December). Corporate Governance and Agency Conflicts. Journal of Accounting Research,
Vol. 46(No. 5), pp. 1143-1181.
Dichev, I., Graham, J., Cambell, H., & Rajgopal, S. (2013). Earnings quality: Evidence from the field.
Journal of Accounting and Economics, Vol. 56, pp. 1-33.
Economist. (2015, January). America’s oil industry: The tough get going. (3), 1-3.
Gervais, S., Heaton, J., & Odean, T. (2011). Overconfidence, Compensation Contracts, and Capital Budgeting. The journal of Finance, Vol. 66( No. 5), pp. 1735-1777.
Graham, H., & Rajgopal, S. (2005). The economic implications of corporate financial reporting. Journal
of Accounting and Economics, 40, pp. 113-142.
Healy, P., & Wahlen, J. (1999). A Review of the Earnings Management Literature and Its Implications for Standard Setting. American Accounting Association, pp. 365-383.
Holthausen, R., Larcker, D., & Sloan, R. (1994). Annual bonus schemesand the manipulation of earnings. Journal of accounting earnings, pp. 29 - 74.
Hribar, P., & Collins, D. (2002, March). Errors in Estimating Accruals: Implications for Empirical Research. Journal of Accounting Research, Vol. 40(No. 1), pp. 105-134.
Jensen, M., & Meckling, W. (1976, July). Theory of the firm: Managerial Behavior, Agency Costs and Ownership Structure. pp.305-360.
33
Jones, J. (1991). Earnings Management During Import Relief Investigations. Journal of accounting
Research, Vol. 29 (No. 2), pp. 193-228.
Kothari, S., Leone, A., & Wasley, C. (2005). Performance matched discretionary accrual measures.
Journal of accounting Economics, 39, pp. 163-197.
Larcker, D., & Tayan, B. (2011). Seven Myth of Corporate Governance. Stanford Closer Look Series, pp. 1 - 9.
Larcker, D., Richardson, S., & Tuna, I. (2007, July). Corporate Governance, Accounting Outcomes, and Organizational Performance. American Accounting Association, Vol. 82(No. 4), pp. 963-1008. Li, L., & Kuo, C.-S. (2017). CEO equity compensation and earnings management: The role of
growthopportunities. pp. 289-295.
Murphy, K., & Zimmerman, J. (1993). Financial performance surrounding CEO turnover*. Journal of
Accounting and Economics, Vol. 16, pp. 273-315.
O.Connor, J., Priem, R., Coombs, J., & Gilley, M. (2006, June). Do CEO Stock Options Prevent or Promote Fraudulent Financial Reporting. The Academy of Management Journal, 49(No. 3), pp. 483 - 500.
Reuters. (2013, January 31). Reserves challenge clouds Shell's growth ambitions. Reuters.
Roychowdhury, S. (2006). Earnings management through real activities manipulation. Journal of
Accounting and Economics(42), pp. 335-370.
Sanders, G. (2001, Jun). Behavioral Responses of CEOs to Stock Ownership and Stock Option Pay.
Academy of Management, Vol. 44(No. 3), pp. 477-492.
Schrand, C., & Zechman, S. (2012). Executive overconfidence and the slippery slope to financial misreporting. Journal of Accounting and Economics, Vol. 53, pp. 311-329.
SEC. (1934). Securities Exchange Act of 1934. Retrieved March 20, 2017, from SEC enforcement accounting and auditing Enforcement releases: https:// www.sec.gov/ divisions/ enforce/ friactions.shtml
SEC. (2004, August 24). Royal Dutch Petroleum Company and The "Shell" Transport and Trading Company, p.l.c./ Lit. Rel. No. 18844 / August 24, 2004. SEC publication.
34
Tarjo, N. (2015). Application of Beneish M-Score Models and Data Mining to Detect Financial Fraud.
ScienceDirect, pp. 924-930.
Taylor, G., & Zhaohui, R. (2010). Consequences of real earnings management on subsequent operating performance. Research in Accounting Regulation, Vol. 22, pp. 128-132.
Thiruvadi, S., & Huang, H.-W. (2011, November 7). Audit committee gender differences and earnings management. International Journal, Vol. 26, pp. 483-498.
U.S. Securities and Exchange Commission. (2017, 03 22). Accounting and auditing enforcement Releases. Retrieved from www.sec.gov: https://www.sec.gov/divisions/enforce/friactions.shtml Zhang, X., Bartol, K., Smith, K., & Pfarrer, M. (2008, April). Earnings Manipulation and Stock-Based
Incentive Misalignment. Academy of Management, Vol. 51(No. 2), pp. 241-258.
Zhao, Y., Chen, K., Zhang, Y., & Davis, M. (2012). Takeover protection and managerial myopia: Evidence from real earnings management. Journal Accounting Public Policy, Vol 31, pp. 109-135.
35
7 Appendices
7.1 Appendix 1: NYMEX daily closing Crude Oil price
Figure 2:
7.2 Appendix 2: Results for the Leven’s homogeneity test
Table 7: Levene’s test
Variable Df, n F Score Conclusion
Oil prices df(2, 1545) Pr > F = 0.0000 Unequal Variance
M-score df(2, 1295) Pr > F = 0.0000 Unequal Variance
Stock ownership df(2, 393) Pr > F = 0.0478 Equal Variance Option ownership df(2, 393) Pr > F = 0.0064 Unequal Variance
CEO duality df(2, 286) Pr > F = 0.1114 Equal Variance
Book-to-Market df(2, 1242) Pr > F = 0.0000 Unequal Variance Return on Assets df(2, 1286) Pr > F = 0.0000 Unequal Variance
36
7.3 Appendix 3: Results of the tests for multicollinearity
In table 8 below the VIF-tests to check for multicollinearity between the variables tested are displayed per model. It seems that multicollinearity can only be expected between CEO duality and the interaction variables since the VIF values are a higher value than 4. Besides Model 5, there does not seem to be multicollinearity.
Table 8: VIF test Variable
Model Model 1 Model 2 Model 3 Model 4
VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF
Control Variables
Book-to-Market 1.02 0.98 1.06 0.94 1.08 0.93 1.09 0.91 Return on Assets 1.01 0.99 1.40 0.71 1.47 0.68 1.57 0.64 CEO duality 1.03 0.97 1.03 0.97 1.05 0.95 1.06 0.94
Main effect variables
Stock ownership 1.68 0.59 2.05 0.49 Option ownership 1.53 0.65 2.05 0.49 Period: post 2014 1.41 0.71 1.43 0.70 19.98 0.05 Interaction variables Stock * Period 17.94 0.06 Option * Period 3.37 0.30 Average VIF 1.02 1.23 1.37 6.14
37
7.4 Appendix 4: SIC industry codes at the two-digit level of the sample firms
Table 9: SIC codes
0100-0999 Agriculture, Forestry and Fishing
1000-1499 Mining (1311 Crude Petroleum & Natural Gas, 1381 Drilling Oil & Gas
Wells, 1382 Oil & Gas Field Exploration Services, 1389 Oil & Gas Field Services)
1500-1799 Construction 1800-1999 not used
2000-3999 Manufacturing (2911, Petroleum Refining)
4000-4999 Transportation, Communications, Electric, Gas and Sanitary service 5000-5199 Wholesale Trade
5200-5999 Retail Trade
6000-6799 Finance, Insurance and Real Estate 7000-8999 Services
9100-9729 Public Administration 9900-9999 Nonclassifiable
38
7.5 Appendix 5: Firms observed with a negative gross margin
Table 10: Listing of firms adjusted for negative gross margin
# Company Name gvkey ticker Net sales in millions GM %
2014 2015 2016 2014 2015 2016
1 APACHE CORP 001678 APA 13.749 6.383 5.367 38,7% -337,4% 57,1% 2 APPROACH RESOURCES INC 178798 AREX 259 131 81,2% -94,0%
3 ATHABASCA OIL CORP 184651 ATHOF 115 175 -42,9% -442,8%
4 ATLAS RESOURCE PARTNERS LP 170781 ARPJQ 682 740 -33,2% -65,1%
5 BAYTEX ENERGY CORP 029135 BTE 1.530 888 602 65,8% -31,9% -18,9%
6 BELLATRIX EXPLORATION LTD 108391 BXE 469 320 200 68,6% -102,5% 169,3% 7 BILL BARRETT CORP 162459 BBG 472 207 179 69,8% -199,0% 77,1% 8 BONANZA CREEK ENERGY INC 187698 BCEI 559 293 195 48,0% -185,5% 58,1% 9 BONAVISTA ENERGY CORP 026796 BNPUF 1.094 622 338 46,2% -69,5% 32,4% 10 BREITBURN ENERGY PARTNERS LP 174656 BBEPQ 1.430 1.109 469 64,7% -155,5% -38,0%
11 CALIFORNIA RESOURCES CORP 021431 CRC 4.173 2.403 1.547 0,2% -152,1% 33,9% 12 CALLON PETROLEUM CO/DE 015060 CPE 152 138 201 78,9% -78,4% 25,5% 13 CANACOL ENERGY LTD 107559 CNNEF 222 186 199 58,9% 7,6% 53,1% 14 CARDINAL ENERGY LTD 019359 CRLFF 221 193 127 65,5% -36,5% -7,4%
15 CARRIZO OIL & GAS INC 065220 CRZO 710 429 444 84,2% -212,9% -58,0%
16 CHAPARRAL ENERGY INC 165556 0957B 682 324 252 73,9% -399,6% -54,7%
17 CHESAPEAKE ENERGY CORP 027786 CHK 20.951 12.764 7.872 32,7% -124,3% -26,8%
18 CIMAREX ENERGY CO 150699 XEC 2.681 1.597 1.380 63,7% -180,8% -0,2%
19 COMSTOCK RESOURCES INC 023002 CRK 555 252 176 71,7% -252,8% 45,6% 20 CONCHO RESOURCES INC 177884 CXO 2.660 1.804 1.635 63,0% 66,6% -20,9%
21 CONTANGO OIL & GAS CO 022053 MCF 276 117 78,8% -163,9%
22 CREW ENERGY INC 156093 CWEGF 373 149 151 6,5% 11,2% 19,8% 23 DENBURY RESOURCES INC 020653 DNR 2.417 1.244 961 62,5% -351,2% -42,1%
24 DEVON ENERGY CORP 014934 DVN 19.566 13.145 10.304 50,5% -114,4% -11,1%
25 DIAMONDBACK ENERGY INC 170750 FANG 496 447 527 81,6% -109,6% 29,0% 26 ECLIPSE RESOURCES CORP 020513 ECR 138 255 235 50,2% -224,6% 34,8% 27 ENCANA CORP 011781 ECA 8.019 4.422 2.918 55,6% -101,6% -22,7%
28 ENERGY XXI LTD 282189 EXXIQ 1.231 1.405 67,5% -107,4%
29 ENERPLUS CORP 020214 ERF 1.761 1.027 693 69,8% -80,9% 0,0%
30 EOG RESOURCES INC 016478 EOG 17.474 8.718 7.363 77,9% -2,4% 69,0% 31 EV ENERGY PARTNERS LP 175006 EVEP 339 178 185 33,8% -9,7% -9,1%
32 EXCO RESOURCES INC 007422 XCO 660 328 271 70,3% -323,7% -25,8%
33 GEOPARK LTD 278151 GPRK 429 210 193 69,6% -12,7% 68,0% 34 GRAN TIERRA ENERGY INC 164046 GTE 559 276 289 32,2% -60,7% -155,3%
35 GULFPORT ENERGY CORP 026069 GPOR 671 709 386 79,0% -134,6% -149,7%
39
TABLE 10 CONTINUED gvkey Ticker 2014 2015 2016 2014 2015 2016
37 HUSKY ENERGY INC 013994 HUSKF 24.092 16.369 12.919 23,8% -6,4% 24,0% 38 ITHACA ENERGY INC 174474 IACAF 439 286 193 -75,9% -141,2% 43,7% 39 JOURNEY ENERGY INC 020918 JRNGF 173 106 -27,1% -92,7%
40 LAREDO PETROLEUM INC 187961 LPI 794 607 597 73,1% -345,3% 26,0% 41 LEGACY RESERVES LP 166436 LGCY 670 437 273 -1,3% -93,3% 6,5% 42 LIGHTSTREAM RESOURCES LTD 183667 LSTMF 1.020 449 9,1% -134,5%
43 LINN ENERGY INC 164109 LNGG 4.983 2.883 25,7% -107,8%
44 LONG RUN EXPLORATION LTD 184075 WFREF 616 327 6,2% -56,1%
45 MATADOR RESOURCES CO 187812 MTDR 431 316 264 80,4% -183,0% 0,4% 46 MEMORIAL PRODUCTION PRTRS LP 194755 MEMP 494 358 -16,5% -136,1%
47 MEMORIAL RESOURCE DEV CORP 020168 MRD 899 732 28,7% -34,3%
48 MURPHY OIL CORP 007620 MUR 5.289 2.787 1.810 76,6% -21,7% 61,4% 49 NEWFIELD EXPLORATION CO 029173 NFX 2.288 1.557 1.472 72,9% -250,5% -7,9%
50 NORTHERN OIL & GAS INC 142337 NOG 595 275 145 83,3% -350,0% -105,8%
51 OCCIDENTAL PETROLEUM CORP 008068 OXY 19.312 12.480 10.090 26,6% -23,0% 41,1% 52 PACIFIC EXPLORTN & PROD CORP 105222 PEGFF 5.743 3.909 1.895 36,0% -112,2% 15,6% 53 PARAMOUNT RESOURCES LTD 020676 PRMRF 335 419 247 50,7% -43,7% 79,3% 54 PBF ENERGY INC 196159 PBF 19.828 13.124 15.920 2,1% 4,8% 5,0% 55 PENGROWTH ENERGY CORP 020683 PGH 1.633 1.018 487 39,4% -20,6% 36,6% 56 PENN WEST PETROLEUM LTD 020684 PWE 2.068 1.265 575 31,3% -85,7% -7,7%
57 QEP RESOURCES INC 154357 QEP 3.414 2.019 1.377 17,9% 34,6% -37,4%
58 RESOLUTE ENERGY CORP 179782 REN 329 155 164 18,1% -420,9% -5,1%
59 REX ENERGY CORP 177896 REXX 305 228 67,1% -71,7%
60 SANCHEZ ENERGY CORP 188663 SN 666 476 431 48,2% -225,4% 18,1% 61 SANDRIDGE ENERGY INC 176899 SD 1.559 769 392 56,6% -538,7% -207,3%
62 SANTOS LTD 100165 SSLTY 3.330 2.400 2.627 24,5% -42,4% -11,1%
63 SAVANNA ENERGY SVCS CORP 136266 SVGYF 792 446 324 -21,3% 1,9% 23,7% 64 SRC ENERGY INC 178870 SRCI 104 125 107 83,1% 66,1% -125,4%
65 STONE ENERGY CORP 028564 SGY 796 545 22,6% -180,9%
66 SURGE ENERGY INC 114211 ZPTAF 444 194 146 28,9% -66,3% 48,7% 67 TRICAN WELL SERVICE LTD 064107 TOLWF 2.704 1.188 325 14,2% 0,1% -0,8%
68 TRILOGY ENERGY CORP 163697 TETZF 549 268 182 69,3% -0,3% 50,1% 69 ULTRA PETROLEUM CORP 108645 UPL 1.182 986 721 69,7% -256,7% 60,3% 70 UNIT CORP 010877 UNT 1.573 854 602 41,0% -148,0% 15,7%
71 VAALCO ENERGY INC 027199 EGY 128 -1,9%
72 VANGUARD NATURAL RESOURCES 178684 VNRSQ 788 567 345 45,6% -258,2% -101,0%
73 W&T OFFSHORE INC 160341 WTI 949 507 400 69,2% -136,6% -14,1%
74 WHITECAP RESOURCES INC 185230 SPGYF 878 482 536 83,2% -14,2% 120,5% 75 WHITING PETROLEUM CORP 155393 WLL 3.055 2.109 1.300 54,9% -9,1% 61,2% 76 WPX ENERGY INC 187128 WPX 3.493 1.888 693 51,9% -66,0% 27,7%
