CEO stock option pay and organizational risk-taking behavior in the post-sox era

(1)

MSc Thesis:

CEO STOCK OPTION PAY AND ORGANIZATIONAL

RISK-TAKING BEHAVIOR IN THE POST-SOX ERA

Martijn Geerdink (10670068) 21st of June 2015

Word count: 12,512

MSc Accountancy & Control, variant Control Amsterdam Business School

Faculty of Economics and Business, University of Amsterdam Supervisor: Prof. dr. V.S. Maas

(2)

2 Statement of Originality

This document is written by Student Martijn Geerdink who declares to take full responsibility for the contents of this document.

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

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

(3)

3

ABSTRACT

Building on Agency Theory, this study provides empirical evidence on the relation between CEO stock option pay and organizational risk-taking behavior. The concept of risk-taking behavior is unpacked by translating the phenomenon into two variables, the degree of R&D investment intensity and stock price volatility. Hypotheses regarding the expected relation are tested on causality. In doing so, this study accounts for an increased reliability of corporate financial reporting by focusing on a sample of firms which are listed in the S&P 500 index in the post-SOX period. The results confirm prior research by finding that an increase in the degree of stock option pay, as a component of total CEO remuneration, increases organizational risk-taking behavior. Finally, insights on remuneration policies are provided from an agent’s perspective.

KEY WORDS: CEO remuneration; Stock options; Incentives; Risk-taking behavior; Agency theory, SOX

(4)

4

1. INTRODUCTION

1.1 Background

In recent decades, multiple corporate financial scandals have occurred due to fraudulent accounting practices and inadequate monitoring practices. The Sarbanes-Oxley Act of 2002 (henceforth, SOX) has been introduced, and studies have been conducted to reveal the fundamental causes resulting in these remarkable phenomena. The aim of this thesis is to scrutinize the role of incentives in risk-taking behavior shown by large corporations through constructing a deeply-rooted theoretical framework and conducting an empirical archival

research. The thesis builds on agency theory, taking both the perspective of the principal and the agent in the large corporation setting.

Large corporations are often characterized by a separation of ownership and control (Zeitlin, 1974). This phenomenon divides the historical entrepreneur in two parties, one party is represented by the manager, the other party is represented by the bearer of risks (Jensen & Meckling, 1976; Fama, 1980). By nature, this division creates a misalignment of interests and an asymmetry of possession of knowledge. Executives, or the ‘ agents’, run the company and owners, or the ‘principals’, try to monitor the executives. Two basic assumptions of economic models state that, by nature, managers are greedy and managers are risk-averse. In order to align the interests between principals and agents and create goal congruence, the most popular tools used are compensation and incentive schemes, together forming the total compensation. Managers are rewarded for their fulfilling their tasks by compensation, and are triggered to create additional value for the firm and shareholders by incentives. In addition, other agency costs are incurred to assure interest alignment to some degree (Jensen & Meckling, 1976).

Agency conflicts and related incentive schemes for interest alignment can have various potential outcomes for corporations. In the recent past, corporate financial scandals like Enron and WorldCom were fueled by the greedy nature of key executives who were concerned with boosting the stock prices. These examples indicate that CEOs (and other executives) are willing to take risks to satisfy their needs despite of their nature of being risk-averse. The debate on determinants for risk-taking behavior has flared up since the occurrence of these scandals. Focusing on incentives triggering risky CEO behavior which aim to create an alignment of

(6)

6 interests between principals and agents, it is very interesting to consider the implications for corporations.

Recent research has shown that there are various factors driving or constraining risky behavior shown by CEOs. A CEO’s strategic and financing decisions, and related attitude towards risk are influenced by personal life experiences, overconfidence, and leverage

preferences (Cronqvist et al., 2012; Malmendier & Nagel, 2011; Malmendier et al., 2011). In his research, Serfling (2014) finds that there is a significant relationship between CEO age and risk-taking behavior and firm performance. Among his findings can be found that older CEOs manage firms with lower stock return volatility and older CEOs invest less in R&D, diversify their firms, and lower operating leverage. Sanders & Hambrick (2007) elaborate on the concept of managerial risk taking and define three major elements, the size of an outlay, the variance of potential outcomes, and the likelihood of extreme loss. The authors build on the agency theory implying that managers tend to be more risk-averse than shareholders would prefer managers to be (Eisenhardt, 1989). A reason underlying this phenomenon is that managers and their

reputations are closely linked to the corporation they work for that taking risks may have crucial downside outcomes (Sullivan & Spong, 2007). The principals (shareholders), on the other hand, are seeking to maximize returns for their investments. These shareholders tend to have a

diversified portfolio of shares and seek to have high returns, which go along with taking big risks by the CEO(Core, Guay, & Larcker, 2003). In order to encourage CEOs to extend their risk-taking behavior, monetary incentives are put into practice, serving as a control system.

Studies have indicated that stock options function as an incentive which increases risk-taking behavior (Rajgopal & Shevlin, 2002; Sanders & Hambrick, 2007). These stock options are lucrative for CEOs, shifting their interests from the prevention of downside losses to the

generation of upside gains. Rajgopal & Shevlin (2002) limit their research to a sample of oil and gas producers, constraining the generalizability of the results. The research conducted by Sanders & Hambrick (2007) contributes by stating that CEOs with high amounts of stock options tend to deliver more big losses than big gains for the company. In their research, the authors included firms listed in the S&P 500, Mid-Cap, and Small-Cap indexes with data ranging from 1993 to 2000. This period is characterized as one with a major burst in CEO stock option pay and the failure of corporate governance systems, leading to overvalued equity and consequently corporate financial scandals (Jensen & Murphy, 2004). The sample included in the study

(7)

7 represents a cross-sectional group of companies based on firm size. Where previous mentioned studies covered research samples and periods which left serious opportunities for earnings management, this thesis focuses on the relation between CEO stock option pay and

organizational risk-taking behavior in a more stable time frame (post-SOX and pre-Global Financial Crisis, 2003 -2007), influenced by the SOX legislation and therefore representing increased reliability of financial reporting practices and subsequent valuation of equity,

contributing to the generalizability of the results . Moreover, the sample included in the analysis is comprised solely of firms listed in the S&P 500. The chosen sample provides generalizability as they represent one of the purest forms of division of ownership and control, complying with the agency theory. CEO age and firm size will be held into account while conducting the research to assure validity of the results.

1.2 Research Question and Contribution

Stock options are incentives used by principals to align the interests of the agents, in this case create shareholder value. But as the stock price is a reflection of how a company is expected to perform in the future, financially and non-financially, it is not very clear how stock options influence organizational risk-taking behavior after the introduction of enhanced legislation and the related decrease in earnings management. In the investigation of the fundamental relations guiding this study, a direct effect of CEO stock option pay on two indicators of organizational risk-taking, R&D investments and stock price volatility, is examined. Therefore, my research question is formulated as the following:

“How does CEO stock option pay influence organizational risk-taking behavior?”

From a high-level perspective, I test how variable compensation affects risk-taking behavior. Risk taking behavior is operationalized as (1) R&D expense over total expense and (2) annual volatility of the stock price. Variable compensation is operationalized as stock option

compensation over total compensation. Since I assume that risk-taking behavior and its measures are caused by the compensation policy in the previous year, the compensation variables are lagged. The relationships are tested across firms (𝑖) and time (𝑡) yielding the following economic models:

(8)

8 𝑅𝐷 𝐸𝑥𝑝𝑒𝑛𝑠𝑒_𝑖,𝑡 𝑇𝑜𝑡𝑎𝑙 𝐸𝑥𝑝𝑒𝑛𝑠𝑒_𝑖,𝑡 = 𝛼 + 𝛽 × 𝑜𝑝𝑡𝑖𝑜𝑛 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛_{𝑖,𝑡−1} 𝑡𝑜𝑡𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛_{𝑖,𝑡−1} + 𝛾 × 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑖,𝑡+ 𝜖 𝑎𝑛𝑛𝑢𝑎𝑙 𝑣𝑜𝑙𝑎𝑡𝑖𝑙𝑖𝑡𝑦_𝑖,𝑡 = 𝛼 + 𝛽 ×𝑜𝑝𝑡𝑖𝑜𝑛 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1 𝑡𝑜𝑡𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛_{𝑖,𝑡−1} + 𝛾 × 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑖,𝑡+ 𝜖

Appropriate control variables are CEO age because older CEOs tend to decrease organizational risk-taking (Serfling, 2014) and firm size because it is assumed to affect investment levels (Sanders & Hambrick, 2007). Based upon prior literature, I expect a positive relationship between CEO stock option pay and organizational risk-taking behavior.

Taking on an agency theory perspective, this thesis makes an empirical contribution to existing literature by finding a coherence between the alignment of interests between principal and agent set by stock option grants, and CEO behavior expressed in terms of organizational risk-taking in a post-SOX era. From a societal perspective, this thesis contributes by proposing a contingency approach for organization-dependent risk preferences and corresponding CEO remuneration practices. This thesis will have value for investors in determining their securities portfolio allocation, for remuneration committees in aligning CEO remuneration policies with the firm’s attitude towards risk, and for shareholders in assigning the Board of Directors and related remuneration committee.

1.3 Outline

This thesis proceeds as follows. In the next section relevant literature is presented and discussed, followed by hypotheses. The succeeding section provides a methodology for conducting the empirical analysis and tests the hypotheses by means of regression analyses. The final section discusses the results and states conclusions, recommendations, and limitations. Last-mentioned section ends with pointing out a direction for future research.

2. LITERATURE REVIEW

In this section of the thesis the theoretical construct and main literature is elaborated upon. First a thorough review of prior literature on agency theory is given to provide the reader with a deep

(9)

9 understanding of the underlying theory concerning the research topic. Second, managerial self-interest is discussed revealing the nature of CEOs and underlying reasons for CEO financial and strategic decision making. Third, risk-taking behavior by CEOs and organizations is debated upon. Fourth, CEO remuneration is discussed covering the components of total remuneration and the role of stock options. Moreover, stock option pay escalation in the 90s and corporate

governance failure are highlighted in perspective of the corporate financial scandals occurring in the early years of the new millennium. Fifth, the responding introduction of the new legislation, the Sarbanes-Oxley Act of 2002, is presented and discussed. Finally, at the end of this section, the indicators of risk-taking behavior are determined, resulting hypotheses are stated, and a conceptual model is presented to illustrate the relations expected.

2.1 Agency Theory

Two authors who proposed a theory of agency in its beginning fase, independently from one another, were Ross (1973) and Mitnick (1973). In his article, Ross develops an economic theory of agency in which agency is seen as a problem resulting from compensation contracting, or differently stated, an incentives problem. Principals strive to select an appropriate incentive mechanism in order to produce the desired behavior of the principle by the agent. Mitnick (1973) builds on an institutional theory of agency, stating that society creates institutions addressing problems which arise in the relationships between the principal and the agent. These different approaches to the agency problem give unique insights in understanding the phenomenon and can be used complementary.

Several years later, Jensen & Meckling (1976) proposed an agency theory of the firm. The authors integrate the theory of agency, the theory of property rights and the theory of finance into one theory, the ownership structure of the firm. The firm is a set of contracts among factors of production in which each factor is motivated by self-interest (Jensen & Meckling, 1976; Fama, 1980). Fama (1980) states that the firm is disciplined by competition from other firms which enables the monitoring of the entire firm performance. Moreover, these markets provide opportunities for managers as they are a factor of production. Therefore, incentives also act as a tool to retain managers.

Jensen & Meckling step aside from the concept of the entrepreneur and divide the traditional functions of the entrepreneur into separate factors, namely management and risk

(10)

10 bearing. Based on the definition of the firm as provided by Alchian & Demsetz (1972), Jensen & Meckling address the agent of modern corporations as the “one party who is common to all the

contracts of joint inputs” and the one “who has the right to renegotiate any input’s contract independently of contracts with other input owners”. The principal of the modern corporation is

addressed as the one “who holds the residual claim” and “who has the right to sell his central

contractual residual status”. With this separation of ownership (or the bearing of risk) and

control (or management) a contract is needed under which the person in control (the agent) performs tasks on behalf of the owner (the principal). Jensen & Meckling (1976) argue that both parties to the relationship are utility maximizers and therefore it is assumable that the agent will not always act in the best interests of the principal. In order for the principal to provide itself with some degree of assurance that the agent will act in the principal’s best interest, incentives are set into practice for the agent. In addition, the principal will incur costs to monitor or restrict actions of the agent, called monitoring costs, the principal will incur costs to make sure the agent is committed to contractual obligations, called bonding costs, and the principal will incur costs related to the reduction in welfare of the principal caused by deviating interests of the principal and the agent even though monitoring and bonding is put into practice, called the residual loss. Together, these incurred costs constitute the agency costs according to Jensen & Meckling (1976). The allocation of these costs is dependent on the degree a principal wants to control the agent in their behavior and outcome. This trade-off aims to result in the optimal contract between the principal and the agent (Eisenhardt, 1989). This thesis focuses on the incentives part of the agency costs in which the principals try to grant the optimal level of stock options to CEOs in order to create shareholder value.

Gottschalg & Zollo (2007) shed some light on the relation between interest alignment and competitive advantage. The authors elaborate on different factors which influence human

motivation and subsequently competitive advantage. The paper by Gottschalg & Zollo (2007) may give some important insights on how stock options function in the alignment of interests between principals and agents. There is reason to assume the optimal contract and an appropriate stock option incentive scheme result in competitive advantage and thereby provide the firm with high financial performance. In case of excessive granting of stock options to CEOs, the interest alignment may be distorted, and may eventually result in poor financial performance. Sanders &

(11)

11 Hambrick (2007) support this possible outcome by arguing, based on their findings, that the heavy use of CEO stock option pay often generates more excessive losses than excessive profits.

2.2 Managerial self-interest

In order to obtain a better insight in the underlying reasons for the necessity of incentives and the incurring of monitoring costs, it is important to gain some knowledge on managerial self-interest. As previously stated, each factor of production, together forming the firm, is motivated by self-interest (Jensen & Meckling 1976). Empirical evidence supporting the nature of self-self-interest held by factors of production is found in a case study (Bouwens & Kroos, 2010) on store managers of the Free Record Shop, a retail chain, who decrease sales activities at crucial moments regarding their compensation conditions. In their research, Bouwens & Kroos (2010) conducted a case study in which they examined how managers respond to bonus schemes. When a manager achieves the quarterly sales target, not set by the manager himself, he or she receives a bonus. The sales target for the next year is based on current year’s sales performance. Bouwens & Kroos (2010) find that the managers in the case study reduce their end-of-year sales

performance, when already achieved the targets of the first three quarters of the fiscal year, in an attempt to mitigate the increase in their next-year sales target. Note that in this case, managers are not able to influence accounting practices in order to influence results. Managers can solely influence sales by engaging actively or less actively in sales activities. Thus, while it would be in best interest of the firm and shareholders to maximize the end-of-year sales performance, managers choose to act in favor of themselves.

As previously mentioned, CEO’s economic means and reputations are closely linked to their firm. Therefore, CEOs create a vast aversion to downside outcomes related to failing by the firm and tend to be risk averse (Sanders & Hambrick, 2007). Serfling (2014) and Cassell et al (2012) give an indication that CEOs are risk-averse by nature. Older CEOs, close to retirement, tend to take fewer risks than young CEOs as pension benefits and deferred compensation come into play (Serfling, 2014). This indicates that CEOs make decisions and take risks based on self-interest.

Watts & Zimmerman (1978) proclaim that managers are self-interested and therefore make choices to maximize their own wealth. This may explain that CEOs tend to be risk-averse, avoiding downward outcomes of the firm, but when stock options come into play CEOs change

(12)

12 their behavior and make financial and strategic decisions which have a higher chance of fueling the stock price. These decisions can be characterized as risky based on the size of an outlay, the variance of potential outcomes, and the likelihood of extreme loss (Sanders & Hambrick, 2007).

2.3 Risk-taking behavior by CEOs

As the principle-agent literature focuses on determining the optimal contract based on behavior versus outcome (Eisenhardt, 1989), the contract between principal and agent is associated with the transferring of risks from the principal to the agent. A contract based on behavior has a stronger guidance on what actions a CEO should take in order to receive his or her

compensation. A contract based on outcomes gives more autonomy to the CEO, and thereby transfers the risks to the CEO who should take decisions in order to achieve the desired results. CEO risk-taking translates to organizational risk-taking as these executives are in such a position that grants them the power to directly influence strategic and financial decision making of the company in question. Therefore, regarding organizational risk-taking, the focus of explaining the concepts is on the underlying factors and reasons for CEO risk-taking behavior.

Following Eisenhardt (1989), CEOs tend to be more risk-averse than shareholders would prefer them to be. As previously mentioned in this paper, Sanders & Hambrick (2007) support the statement made by Eisenhardt (1989), arguing that CEOs have a great part of their financial resources as well as their reputations tied to the firm they manage. This creates a strong dislike of CEOs towards downside outcomes. On the principal side of the contract, the

shareholders are risk-neutral as they generally have a widely diversified portfolio of shares. The main goal of shareholders is maximizing returns on their investments. With this asymmetry of interests between CEOs and shareholders, alignment is needed to achieve an increased goal congruence. In their article, Hall & Liebman (1998) state that: “The most direct solution to (the)

agency problem is to align the incentives of executives with the interests of shareholders by granting (or selling) stock and stock options to the CEO”. This statement proclaims that by

means of stocks and stock options, managers will tend to act more in the best interest of the shareholders. Thereby, shareholder returns are aimed to be maximized.

What is not taken into account in the statement is that by loading CEOs with stock options, behavior is influenced in such a way that CEOs will take more risks which increases the potential for excessive downside outcomes and the potential for excessive positive economic

(13)

13 outcomes. This can be seen in the stock price volatility resulting from risky CEO behavior as managers are self-interested and make choices for own wealth maximization (Watts &

Zimmerman, 1978). These excessive downside outcomes caused by CEO behavior can have a massive impact on shareholder value, but not for the personal financial wealth of CEO of the company. Sanders & Hambrick (2007) explain the increasing risk-taking behaviors caused by stock options by stressing the consequences. The authors argue that the granting of stock options to CEOs encourage CEOs to invest financial assets in uncertain categories, such as research and development (R&D), capital expenditures, and acquisitions. Furthermore, they find statistical support for their hypothesis, that states that the heavy use of CEO stock options brings about more big losses than big wins. This relation is supported by the argument that heavy stock option loaded CEOs have huge personal upside gains in perspective related to increasing performance of the firm, while having no losses in perspective related to low performance of the firm. The latter can be clarified by the fact that CEOs simply have the option not to exercise the stock options. To form a deeper understanding of the relation between stock options and risk-taking behavior, it is essential to put stock options into perspective of total CEO remuneration.

2.4 CEO Remuneration

In order to create long-run shareholder value, the right CEO should be attracted at the lowest possible cost, the right CEO should be retained at the lowest possible cost, and CEO’s behavior should be motivated in an appropriate way (Jensen & Murphy, 2004). A well-designed

remuneration policy, dependent on three dimensions as stated by Jensen & Murphy (2004), is critical to accomplish this. The first dimension is the expected total benefits associated with the job or position (including the costs and benefits of non-pecuniary aspects of the job). This entails that executives will be in the function which has the highest expected total benefits associated with the job. Organizations determine their remuneration policy in such a way that the desired CEOs are attracted and retained, and the executives who are not preferred are encouraged to leave. The second dimension is the composition of the remuneration package. This dimension involves the separate elements of the remuneration package, often including cash payment (salary), stock-options, stock, restricted stock, retirement benefits and non-monetary benefits. The optimal, most efficient remuneration package composition maximizes the CEO’s expected

(14)

14 total benefits and minimizes the company’s total costs. The third dimension is concerned with the relation between pay and performance. This dimension defines which results and which actions taken by CEOs are rewarded or penalized. Therefore, the relation between pay and performance aims to determine what a CEO works on, the extent to which a CEO delivers hard work, and the productivity delivered by the CEO.

These dimensions (Jensen & Murphy, 2004) should be assessed and defined firm-specifically, taking a contingency approach to firm characteristics and attitude towards risk. In addition, the authors stress that the dimensions have to be managed by the remuneration committee consisting of directors who function with high integrity. Thus, a strong corporate governance structure is essential for effective remuneration and the mitigation of agency problems (Jensen & Murphy, 2004; Sullivan & Spong, 2007). Firms who are risk-averse will tend to compose a remuneration package with a high proportion of retirement benefits, thereby encouraging CEOs to stay at the firm for a relatively long period. In contrast, firms with are less risk-averse will tend to embrace high bonus opportunities and other performance-related pay, including stock options, in their remuneration package and thereby have a greater chance of attracting CEOs who are less risk-averse, more optimistic, and more confident about their ability to create value (Jensen & Murphy, 2004).

Executive remuneration practices in US firms have experienced major changes throughout the decades before the introduction of the SOX in 2002. The average total

remuneration for CEOs in S&P 500 firms increased from about $850,000 in 1970 to over $14 million in 2000, and fell to $9.4 million in 2002 (Jensen & Murphy, 2004). Of the mentioned average total remuneration, the average value of stock options comprised from about $0 in 1970, to more than $7 million in 2000, and decreased to $4.4 million in 2002. Note that the share of stock options in total remuneration in 2002, despite the severe cut in total remuneration, still accounts for almost half of total CEO pay. Considering these decades, especially the 1990s have experienced a major burst in CEO pay, mainly driven by the increase of stock option grants. This escalation of stock option is explained as the increasing focus on equity-based compensation by shareholder groups (Jensen and Murphy, 1990a; Jensen and Murphy, 1990b). Another

explanation, supported by Hall and Murphy (2003), highlights the impact of new disclosure rules in the early 90s which require firms to report the number of options granted instead of the value of the options.

(15)

15 The escalation of stock option compensation has led to overvalued equity. Jensen & Murphy (2004) define equity as overvalued when : “a firm’s stock price is higher than its

underlying value. A firm whose stock is substantially overvalued will not be able to deliver the performance the market expects to justify that valuation.”. The higher valuation of equity is an

agency problem, characterized by Jensen (2002) as “agency costs of overvalued equity”. Executives possess knowledge that the shareholders do not know about and present misleading data. These executives, being aware of the fact that they cannot generate the performance needed to support the high stock price of the firm, engage in activities even destroying more of the core value of the firm. Future revenues are shifted to the present and present expenses are shifted to the future by means of accounting practices, and overvalued equity is used to finance

acquisitions (Jensen & Murphy, 2004). These actions postpone the inevitable moment that the company has to admit that it cannot live up to the value as presented by the stock price. In the early years of the new millennium, stock option-loaded executives have often engaged in fraud and manipulation, attempting to preserve the stock prices for even a longer time. Eventually, these practices have resulted in corporate scandals like the well-known Enron case and the WorldCom case. In the US, these scandals have ringed the alarm bells and created a sense of urge for new regulations concerning financial disclosures. The SOX of 2002 is one of the responses.

2.5 The Sarbanes-Oxley Act of 2002

The SOX passed US Congress in July 2002, and has been considered the most far-reaching securities legislation since the Securities Acts of 1933 and 1934 (Zhang, 2007). The US government wanted to restore the public confidence in US capital markets by protecting

investors through improving the accuracy and reliability of corporate disclosures. In addition to the extensive disclosure requirements included in SOX, substantive corporate governance mandates are proposed by SOX, a new phenomenon in the history of securities legislation (Romano, 2004). SOX aims to prevent fraudulent accounting and management misbehavior by demanding more oversight responsibilities, striking harsher penalties for managerial misconduct, and anticipating scandals by dealing with potential conflicts (Zhang, 2007).

(16)

16 Section 404 of SOX deals with the Effective Internal Controls over Financial Reporting (ICFR) required to transparent and reliable financial statements. Section 404 requires

management to set up a report assessing and concluding on the effectiveness of their firm’s internal controls. The external auditor, who is required to be independent of the firm, is

responsible for issuing a report on the effectiveness of internal controls over financial reporting and a verification report on management’s assessment of internal controls. In addition, the Public Company Accounting Oversight Board’s (PCAOB) Auditing Standard No. 2 requires both the independent external auditor and management of the firm to conclude that ICFR is not effective in the case of the existence of material internal control weakness (Chan et al., 2008). Chan et al. (2008) describe material weakness as a flaw that exists “when the design or operation of internal

controls does not allow for the prevention or detection of a misstatement on a timely basis and can likely result in a material misstatement in the interim or annual financial statements”. These

material weaknesses can be the result of weak corporate governance and allow some space for earnings management, which can eventually cause an inaccurate stock price of the firm

2.6 Indicators of organizational risk-taking behavior and hypotheses formulation The first indicator of organizational risk-taking behavior is R&D investments. Following the Financial Accounting Standards Board (FASB), R&D investments include all expenses which can be related to the following definition:

“ Research is planned search or critical investigation aimed at discovery of new knowledge with the hope that such knowledge will be useful in developing a new product or service or a new process or technique or in bringing about a significant improvement to an existing product or process. Development is the translation of research findings or other

knowledge into a plan or design for a new product or process or for a significant improvement to an existing product or process whether intended for sale or use. It includes the conceptual

formulation, design, and testing of product alternatives, construction of prototypes, and operation of pilot plans.”

With the introduction of SOX a more transparent financial disclosure policy is enforced upon US firms, and less opportunity for earnings management remains. The capitalization of R&D expenses are harder to realize as a result. United States General Accepted Accounting Practices (US GAAP) does not allow the capitalization of internal costs for R&D. Under the

(17)

17 Statement of Financial Accounting Standards No. 2R.12 (FAS 2R.12) , R&D expenditures should be expensed when incurred. This implies all R&D investments made are treated as costs and are recorded on the income statement of a company as expenses. According to Kothari et al. (2002), the reason that US GAAP does not allow the capitalization of costs related to the

research nor the development activities, is assumed to be the standard setter’s perceived degree of uncertainty about future economic benefits from current research and development outlays. This uncertainty of investing organizational financial funds in projects reflects risk-taking behavior. There has been an extensive amount of research confirming the coherence of R&D expenditures with organizational risk-taking (Singh, 1986; Coles et al., 2006; Sanders & Hambrick, 2007). CEOs, the agents in the principal-agent relation, often have the power to directly influence organizational decisions with the practical results, such as the scope of R&D investments. To reflect incentives for CEOs to fuel the stock price and are able to do so by means of R&D activities, the following hypothesis is formulated:

H1a: CEO stock option pay is positively associated with R&D investments.

The second indicator of organizational risk taking is the stock price volatility. For this study, stock price volatility is defined as the standard deviation (SD) of the daily stock returns. This definition will be elaborated upon more specifically in the methodology section of this thesis. This indicator too, has been used extensively to reflect risk-taking behavior (Olsen, 1998; Cohen et al., 2000; Sanders & Hambrick, 2007; Cassell et al., 2012). Stock price volatility is caused by public information and private information which affect prices when informed investors trade (French and Roll, 1986). The actual stock price is set by what investors are willing to pay for a security, and what the owners of the security are willing to sell it for, based financial and non-financial expectations of performance of the firm. Investors seek to maximize their economic wealth by investing in securities with high expected returns. When information is disclosed to the public or certain investors, traders will respond depending on the content of the information. Assuming that if stock markets are efficient in the strong-form sense, then any change in long-run shareholder value will be directly reflected by the change in the stock price (Jensen, 1969). Concerning the content of the information ,CEOs are in the position to disclose information about the corporation in question. Therefore, it may be assumed that CEOs can, to some extent,

(18)

18 selectively disclose information directly affecting stock prices (Jensen & Murphy, 2007). E.g. a CEO can disclose information about current R&D projects that have substantial potential of increasing organizational value in the future. Traders can respond optimistically and the stock price will increase. When traders discover that the underlying value of equity cannot live up to the stock price, the stock price will decrease. As CEO stock option pay is associated with the engagement in risky projects, increasing the outcomes of big losses and big gains (Sanders & Hambrick, 2007), the stock price will tend to fluctuate more. To reflect this assumed relation, the following hypothesis is formulated:

H1b: CEO stock option pay is positively associated with stock price volatility

Figure 1: Conceptual model regarding H1a and H1B

3. METHODOLOGY

3.1 Sample

To perform the analysis, the sample in use is a group of US-based firms, the Standard & Poor’s 500 (henceforth, S&P 500). The period in question covers the years 2003 up to and including 2007. The choice for this sample and timeframe is made because of the following arguments. The US is the undisputed trendsetter in executive remuneration practices, being the pioneer on

(19)

19 the introduction of the SOX-reporting legislation. The timeframe concerning this study is post-SOX and pre-crisis. This means that this is a relatively stable time concerning financial reporting and additional organizational reporting, resulting in a relatively true representation of the firm value by the stock prices. These firms are the biggest 500 publically listed firms in the US based on market capitalization. These companies give a reliable representation of developments on the stock exchange market in the US, given the fact that S&P 500 is the most tracked stock exchange index by investors from all over the world. In addition, this sample provides the most accurate availability of data on the firms due to the widespread access to databases. Finally, from an agency theory perspective, these firms give a very reliable representation of large corporations which reflect the most suitable corporations for division of risk bearing and management, or ownership and control.

For the analysis of this research, only the companies who have been listed in the S&P 500 for the entire period of 2003 up to and including 2007 have been selected. Companies who have entered and left the S&P 500 are excluded from the research due to a lack of data for the calculation of the variables. Furthermore, some companies have been lost while the CRSP

database with the Compustat Execucomp database due to missing data on stock price return. This makes it impossible to calculate the stock price volatility for several companies. The actual sample, after deduction of companies with incomplete data, for the analysis consists of 294 unique companies. For these companies, data of a 5 year time frame is examined, resulting in 1,241 unique observations. Note that this number does not add up, as 294 multiplied by 5 equals 1,470 observations. The reason for the smaller number of observations is that some companies do not have the required data for each year, and therefore the observations of these years are excluded from the analysis.

3.2 Data Collection

The data analysis for this thesis is based on archival records which are distracted from accessible databases thru Wharton Research Data Services (WRDS). The database consulted for CEO stock option compensation is Compustat Execucomp. Data for organizational risk-taking behavior is retrieved from Compustat and The Center for Research in Security Prices (CRSP). In order to merge the different databases, firm identification is used. Compustat states company

(20)

20 identification numbers as GVKEY (The Global Company Key). CRSP defines a different

identification convention, PERMNO, which is a permanent identifier of securities. These security identification numbers are linked to the GVKEY by using the linking table of

CRSP/Compustat Merged on WRDS. The result is a single, clean database in STATA which is used for the performance of statistical tests and analysis.

3.3 Dependent Variables

The dependent variable in this study is organizational risk-taking behavior, reflected by R&D investments and stock price volatility. The empirical measures that proxy for this theoretical construct are the percentage of R&D expenses of total expenses and the standard deviation of the daily stock returns of a firm in a given year. Under FAS 2R.12, a formal document issued by the FASB, it is required that all R&D expenses are incurred in the period in question because of the fact that future economic benefits are insecure and severely hard to measure. Therefore, R&D expenses are treated as operating expenses and, in this study, calculated as the percentage of total operating expenses of a company. The first stock price observation of the year is taken to

determine the variance on the return of this stock. The first observation is the closing price of the stock on the 31st December of the previous year. The variance is first calculated on a daily basis, using the stock’s daily return and average value of the stock. From this variance the square root is taken to determine the standard deviation (SD) from the average stock price. When this is done an annualized volatility is calculated and assigned to each company, representing the yearly stock price volatility. The annualized volatility expressed in mathematical terms results in the following formula: 𝜎 = √∑(𝑟𝑖 − 𝑟̅)2 𝑁 − 1 𝑁 𝑖=1 Where:

σ = Annual stock price volatility

N = Number of observations

𝑟̅ = Mean return

(21)

21 3.4 Independent Variable

The independent variable regarding both hypotheses is CEO stock option pay. This variable is operationalized by calculating the percentage of CEO stock option grants of total CEO

compensation. The value of these stock options are based upon the Black-Scholes method, which is calculated by Execucomp. The granting of stock options, as explained in the literature review, is an incentive and therefore is expected to have an influence on decisions made by CEOs in future time periods. This is why I have chosen to use a lagged model for the performance of regression analyses. Following Sanders & Hambrick (2007), I have included two lagged models in my statistical tests. The first model looks at the percentage of stock options of the previous year (t – 1) and its relation to the stock price volatility and R&D investments of the consecutive year. The second model takes an average stock option compensation of the previous three years into account (t – 1, t – 2, and t – 3). Due to the fact that this research focuses on the post-SOX and pre-crisis period, the tests concerning the first model do not give a result for the year 2003 and the tests concerning the second model only show results for years 2006 and 2007.

3.5 Control Variables and Dummy Variables

In this study , control variables were added as they are held constant in order to assess or clarify the relationship between the other variables. The first control variable is firm size. This variable is included to account for the magnitude of R&D investments made by firms and the stock price volatility of the firms in the sample. Firm size is measured by the natural log of revenues. Data on the revenues of the companies in the sample are obtained from Compustat. Serfling (2014) found support for a relation between CEO age and organizational risk-taking behavior.

Therefore, the second control variable is CEO age, measured as the actual age of the CEO in function at years in question. CEO age is obtained from Execucomp.

In the analysis, CEO personal attitudes towards risk based on personal life experiences, overconfidence, and leverage preferences will not be taken into account. This will not be done due to data availability on personal characteristics of CEOs. Therefore, it has to be

(22)

22 acknowledged that the results from this analysis, on ceteris paribus basis, will be limited to some extent. The factors not taken into account may be of considerable importance in determining the degree of CEO risk-taking behavior.

The research will not limit to a specific industry as the S&P 500 index does not limit to firms of a particular industry, resulting in subgroups in the sample. To allow for industry specific variation in the dependent variables, the 2 digit Standard Industrial Classification (SIC) codes function as a dummy variable. This indicator-variable takes the value of 0 or 1 depending on its absence or presence of the particular industry which may influence the outcome. In addition a dummy variable was made for each separate calendar year included in the analysis, also allowing for year-specific variation in the dependent variables.

4. RESULTS

4.1 Preliminary Analyses

Table 1 provides the variable definitions included in this study. I have added the variables which are not in the hypotheses, but are used as a tool to check the validity of the data, under the independent variables. Also, I have added the variables which are used for the calculation of composed variables that could not be distracted from the accessible databases, under the independent variables.

(23)

23

VARIABLES DEFINITION

Dependent Variables

Rd_total The share of R&D expenses expressed as % of total operating expenses StockPrice_Vol Annual stock price volatility (SD of daily stock returns in the given year, %)

Independent Variables

Grant_totalcomp

The share of stock option grants to CEOs expressed as % of total CEO compensation

L.grant_totalcomp

The lagged share of stock option grants to CEOs expressed as % of total CEO compensation at t-1

avgcomp

The average share of stock option grants to CEOs expressed as % of total CEO compensation of t-1, t-2, and t-3

L_Op_Exp The natural log of operating expenses L_R&D_Exp The natural log of R&D expenses

Revenue Total The total amount of revenue (in million USD, $) Operating

Expenses The total amount of operating expenses (in million USD, $) Expenses R&D The total amount of R&D expenses (in million USD, $)

Control Variables

L_Rev_Tot Firm size based on the natural log of revenues

L.logrevt The lagged firm size based on the natural log of revenues at t-1

Age CEO age considering the year in function

Dummy variables

2Dig_SIC Industry type based on 2-digit SIC code

Year Calendar year

Table 1: Variable Definitions

Table 2 presents the descriptive statistics for the main research variables. As can be seen in Table 2, the data derived from several different databases does not represent a balanced panel. The 294 organizations included in the sample, which are usable for the present study, constitute to a total of 1,241 observations. Of the total 1,241 observations, 686 observations have records of

organizational R&D expenses, 894 observations have records of CEO age at the year in function, and 698 observations have data on CEO stock option compensation. For the analysis this entails that, regarding hypothesis H1a, 686 observations are used and regarding hypothesis H1b, 698 observations are considered in the analysis.

(24)

24

VARIABLES N Mean SD Theoretical

Range

Actual Range

Year 1,241 N/A N/A 2003 - 2007 2003 -2007

Revenue Total 1,241 17,749 34,669 0 - ∞ 283.136 - 375,376 Operating Expenses 1,241 14,114 29,417 0 - ∞ 266.2 - 350,592 Expenses R&D 686 588.3 1,148 0 - ∞ 0 - 8,434 Rd_total 686 8.24% 10.7% 0% – 100% 0% - 59.9% StockPrice_Vol 1,241 25.50% 9.30% 0% – 100% 10.79% - 71.81% Age 894 51.92 6.003 0 - ∞ 38 - 80 Grant_totalcomp 698 33.6% 25.7% 0 % – 100% 0 %– 100%

Table 2: Descriptive statistics. Operating Expenses, Expenses R&D, and Revenue Total, and as millions of USD ($). N is the number of observations.

∞ equals infinite

The numbers show that CEO stock option compensation with a mean of 33.6% has a high SD of 25.7%, showing substantial differences between organizational compensation policies. The share of R&D expenses on total operating expenses has a mean of 8.24% with a SD of 10.7%. The magnitude of the SD indicates strongly differing R&D investment behavior by companies in the sample. The annual stock price volatility presents a mean of 25.50%, a SD of 9.30%, and a range with a minimal value of 10.79% and a maximum value of 71.81%. These percentages reflect plausible values, which are in line with possible stock market behavior of the S&P 500 index as well as values presented in comparable studies.

The correlation coefficients in Table 3 shows how the main variables are correlated with each other and if this is a positive or negative correlation, providing a base for estimations of outcomes between variables. The dummy variables year and 2-digit SIC are left out of Table 2 as correlation coefficients are not meaningful for categorical data concerning this study.

(25)

25 VARIABLES L_Rev _Tot L_Op_ Exp L_R&D _Exp Age Grant_to talcomp. StockPric e_Vol Rd_tota l L_Rev_Tot 1.00 L_Op_Exp 0.98 1.00 (0.00)* ** L_R&D_Exp 0.45 0.42 1.00 (0.00)* ** (0.00)*** Age 0.11 0.08 0.01 1.00 (0.00)* ** (0.01)** (0.91) Grant_totalcomp. -0.17 -0.15 0.04 -0.07 1.00 (0.00)* ** (0.00)*** (0.43) (0.17) StockPrice_Vol -0.28 -0.24 -0.04 -0.06 0.22 1.00 (0.00)* ** (0.00)*** (0.40) (0.06)* (0.00)*** Rd_total -0.40 -0.44 0.48 0.02 0.35 0.31 1.00 (0.00)* *** (0.00)*** (0.00)*** (0.64) (0.00)*** (0.00)***

Table 3: Correlation Matrix. The natural log is taken for total revenue (L_Rev_Tot), the operating expenses (L_Op_Exp), and the R&D expenses (L_R&D_Exp). Two-tailed tests performed for all variables.

* Indicates significance at a 10% level (P<0.1) ** Indicates significance at a 5% level (P<0.05) *** Indicates significance at a 1% level (P<0.01)

Considering the independent and dependent variables, the results in Table 3 indicate an assumed correlation. The correlation coefficients show that there is a significant correlation (at P<0.01) between the percentage of stock options granted of total CEO compensation (Grant_totalcomp) and the stock price volatility (StockPrice_Vol), and that these two variables vary with each other positively by 22%. In addition, the correlation coefficients reflect a significant correlation (at P<0.01) between the variables Grant_totalcomp and Rd_total. These variables are indicated to vary together positively for 35%.

I added the additional variables to the correlation matrix to check and confirm for the adequacy and validity of the dataset. The variables added are the natural log of total revenue

(26)

26 (L_Rev_Tot), the natural log of operating expenses (L_Op_Exp), and the natural log of R&D expenditures (L_R&D_Exp). The figures present a positive , almost perfect (98%) correlation between the variable L_Rev_Tot and the variable L_Op_Exp at the significance level of 1%. This indicates, as may be assumed, that bigger firms based on the natural log of total revenues have higher operating expenses in the terms of the natural log of operating expenses. Moreover, contributing to the validity of the data, bigger firms are positively correlated with the natural log of R&D expenses, too significant (at P<0.01). The correlation matrix also tells us that bigger firms have a significant (at P<0.01), negative correlation of 28% with the stock price volatility and a significant (at P<0.01), negative correlation with the percentage of R&D expenses of total expenses of 40%. Furthermore, the control variable CEO age is added to the correlation matrix in order to check assumptions. The results show that CEO age is negatively correlated with the stock price volatility (significant at P<0.1), indicating that as CEOs get older, the stock price volatility will decrease. On the other hand, contrary to assumptions, CEO age is positively correlated with R&D expenses of total expenses. This correlation, however, is very small (2%) and not significant.

To gain some extra insight into the correlation and see whether there is a linear

relationship between the independent variables and the dependent variables, Figure 2 and Figure 3 display a scatterplot using the L.grant_totalcomp, which is the total grant of CEO stock options over the total CEO compensation of the prior year (t-1). The scatterplots do not reveal outliers which should be excluded from the analysis. In both the figures, the independent variable is displayed on the horizontal axis (x-axis) and the dependent variable is displayed on the vertical axis (y-axis). Figure 2 indicates that there is a positive linear relation between the lagged variable of stock option grants to CEOs as part of their total compensation and the R&D expenses as part of the total organizational expenses. The figure presents an extensive amount of dots on the x- and y-axis. It is probable to assume that the reason behind this is that many companies pursue policies by which no stock options are granted as part of CEO total compensation or companies are reluctant to R&D expenses in a given year. Figure 3 also, although less strong, indicates that there is a positive linear relation between the variables. The lagged variable of stock option grants to CEOs as part of their total compensation seem to vary together with the stock price volatility. Again, it is notable that there is an extensive amount of dots on the y-axis, indicating that a lot of companies do not grant stock options. In contrast, there are no dots on the x-axis

(27)

27 simply because no company’s stock price is static.

Figure 2: Scatterplot CEO stock option pay and R&D investments

0 .2 .4 .6 0 .2 .4 .6 .8 1 L.grant_totalcomp

Fitted values rd_total

rd _ to ta l 0 .2 .4 .6 .8 0 .2 .4 .6 .8 1 L.grant_totalcomp

Fitted values StockPrice_Vol

Sto ck Pr ice _ Vo l

(28)

28

Figure 3: Scatterplot CEO stock option pay and stock price volatility

Because the correlation matrix only gives us an insight on how two variables vary with each other and in what direction this correlation exists, additional analyses are needed to determine which relations are present between the variables. The same goes for the scatter plots, correlation does not imply causality. Noteworthy is the fact that the correlation matrix does not represent a lagged model. But although the model is not lagged, the matrix may still be a notable indicator of the cohesion of the independent and dependent variables because companies may be inclined to pursue somewhat similar compensation practices over the years, revealing a company’s attitude towards risk. To answer the research question guiding this study and to test the hypotheses for a predictive relationship, regressions are performed.

4.2 Hypothesis Tests and Robustness Checks

For hypothesis testing, 5 linear regressions are performed for each dependent variable. The results can be found in Table 4 and Table 5. For tests (1) up to and including (4), a lagged model of stock option grants is used and related to the R&D investments or stock price volatility of the upcoming year. In test (5), the average stock option compensation as percentage of total

compensation of the last 3 years is taken (avgcomp). In all 5 tests, the control variables firm size (L_Rev_Tot) and CEO age (Age) are included in the analysis. In tests (1) up to and including (3) the ordinary least squares (OLS) regressions are performed with White robust standard errors. White robust standard errors are added to test whether the results stay while despite the fact that the data is not normally distributed. Tests (4) and (5) perform a cross-sectional time series regression by means of xtregar (tool in STATA).

Regarding hypothesis 1a, stating that CEO stock option pay is positively associated with R&D investments, Table 4 should be consulted. Test (1) presents the results of an OLS test without dummy variables included. The numbers confirm that there is a significant positive relation (at a significance level of 1%) between prior year stock option grants

(L_grant_totalcomp) and R&D investments (rd_total). Test (2) represents the results of an OLS regression with the industry dummies included. Test (3) also includes the year dummies. Both tests result in significant positive relations between L_grant_totalcomp and rd_total at a 1%

(29)

29 significance level. This means that, while allowing for industry- and year-specific variation of these subgroups in the dependent variable, the results stay almost exactly the same. The fact that the results do not alter as an outcome caused by the dummy variables, strengthens the detection of the relation between CEO stock option pay and R&D investments.

Due to the characteristics of the dataset, which is a compound of multiple unbalanced panel datasets and contain first-order autoregressive disturbance (past values have an effect on current values), an OLS regression is not sufficient to adequately test the hypotheses. The results may be misrepresentative to some extent because of magnified t-statistics and understated standard errors. Hence, following Sanders & Hambrick (2007), to fit cross-sectional time series regression models in case of first-order autoregressive disturbance, xtregar is used. By doing so the robustness of the regression is strengthened and proven to a great extent. The xtregar tool serves to regress unbalanced panels of which the observations do not have an equal spread over time. In doing so, the tool offers a generalized least squares (GLS) estimator for the random-effects models. In addition xtregar controls for autocorrelation, thereby preventing bias in the standard errors. The GLS estimator accounts for the unequal variances of the observations in the dataset (heteroskedasticity). The results of xtregar under test (4) adhere to the robustness of the regression concerning hypothesis 1a. A positive significant relation is shown at the 10% level. In the lagged model of test (5), I have altered the measure of the independent variable to the

average stock option compensation as percentage of total compensation of the last 3 years ((t -1 + t – 2 + t – 3) / 3). This test is performed to contribute to the robustness of the regression by assuming that stock options work as an incentive for multiple years thanks to the vesting period attached to the stock options granted , and hence have an effect on the R&D investments. Note that the observations included in the test only cover the years in the post-SOX period and before the financial crisis. Therefore a lagged 3-year average is only applicable in the years 2006 and 2007. Although this leaves us with a rather small amount of observations (93) as usable for the analysis, the result of avgcomp is significantly (at a 1% level) and positively related to rd_total. Based on the previous mentioned test results hypothesis 1a is accepted.

Concerning Table 4 it is remarkable to mention that the control variable CEO age is positively related to R&D investments and is significant in tests (2), (4), and (5). This is in contrast with my assumptions based on the findings presented by Serfling (2014). For the control variable firm size, I have taken the prior year (t-1) natural log of revenue because it is likely to

(30)

30 go along with upcoming year investment decisions. Firm size is negatively related with R&D expenses and significant at a 1% level in the tests (1), (2), (3), and (4). In test (5) this relation is negative and significant at a 5% level.

(1) (2) (3) (4) (5)

VARIABLES rd_total rd_total rd_total rd_total rd_total

L.grant_totalcomp 0.0943*** 0.0919*** 0.0999*** 0.0270* (0.0255) (0.0243) (0.0249) (0.0146) L.logrevt -0.0296*** -0.0272*** -0.0277*** -0.0306*** -0.0226** (0.00549) (0.00640) (0.00642) (0.00751) (0.0109) Age 0.00129 0.00215* 0.00190 0.00231* 0.00398** (0.00112) (0.00121) (0.00117) (0.00131) (0.00191)

Industry dummies Included Included Included Included

Year dummies Included

avgcomp 0.191*** (0.0507) Constant 0.244*** 0.239*** 0.236*** 0.290*** 0.0606 (0.0748) (0.0799) (0.0796) (0.111) (0.170) Observations 256 256 256 256 93 R-squared 0.224 0.511 0.520 Number of gvkey2 100 79

Table 4: The effect of CEO stock option pay on R&D investments. Robust standard errors in parentheses.

Number of gvkey2 is the number of unique companies. * Indicates significance at a 10% level (P<0.1) ** Indicates significance at a 5% level (P<0.05) *** Indicates significance at a 1% level (P<0.01)

Regarding hypothesis 1b, stating that CEO stock option pay is positively associated with stock price volatility, Table 5 should be consulted. Considering test (1), (2), and (3), the relation between the prior year CEO stock option pay (L_grant_totalcomp) and the stock price volatility (StockPrice_Vol) is positive and significant (at a 1% level) based on OLS regression. Test results with and without dummies show almost identical results. Thus, while allowing for variation in the subgroups in the dependent variable, the results strengthen the indication of the presence of

(31)

31 the assumed relation. Just like the regression analysis for hypothesis 1a, White robust standard errors are set in place due to the fact that the data is not distributed normally. The same reasons for making use of xtregar in tests (4) and (5) in the regression analysis of CEO stock option pay and R&D investments, apply for the regression of CEO stock option pay and stock price

volatility. Test (4) shows a positive significant relation between CEO stock option pay and stock price volatility at a 5% significance level. In the lagged model of Test (5), again I have altered the measure of the independent variable to the average stock option compensation as percentage of total compensation of the last 3 years ((t -1 + t – 2 + t – 3) / 3). In this case the result does not show a significant relation between the variables, decreasing the robustness of the regression to some extent. Note that the number of observations in test (5) is rather low, existing of almost one third of the number of observations included in tests (1), (2), (3), and (4). Based on the previous mentioned test results hypothesis 1b is accepted. Furthermore, the control variable firm size is negatively related to stock price volatility. In all tests this is significant at a 1% level. CEO age does not represent a significant relation with stock price volatility.

(1) (2) (3) (4) (5)

VARIABLES StockPrice_Vol StockPrice_Vol StockPrice_Vol StockPrice_Vol StockPrice_Vol

L.grant_totalcomp 0.0473*** 0.0354*** 0.0399*** 0.0282** (0.0135) (0.0130) (0.0133) (0.0117) L.logrevt -0.0201*** -0.0228*** -0.0236*** -0.0184*** -0.0197*** (0.00379) (0.00376) (0.00366) (0.00416) (0.00544) Age -0.000796 0.000169 6.45e-05 0.000163 -0.000497 (0.000643) (0.000566) (0.000541) (0.000739) (0.000932)

Industry dummies Included Included Included Included

Year dummies Included

avgcomp 0.0418 (0.0290) Constant 0.444*** 0.458*** 0.458*** 0.422*** 0.478*** (0.0532) (0.0516) (0.0490) (0.0689) (0.0831) Observations 443 443 443 443 154 R-squared 0.135 0.494 0.515 Number of gvkey2 175 135

Table 5: The effect of CEO stock option pay on stock price volatility. Robust standard errors in parentheses.

(32)

32

* Indicates significance at a 10% level (P<0.1) ** Indicates significance at a 5% level (P<0.05) *** Indicates significance at a 1% level (P<0.01)

5. DISCUSSION

Based on the results from the regression analyses, both hypotheses are accepted. As expected, the independent and dependent variables correlate and relate to each other. The correlation matrix does not represent a lagged model and therefore examines the correlation between the granting of stock options and organizational risk-taking behavior in the same year, somewhat conflicting with the economic models stated in first section of this thesis. But the matrices do indicate a significant correlation between the independent variables and the dependent variables. This may be explained that companies tend to be conservative in their compensation policies and consequently tend to be rather constant in their attitude towards risk. The relation between CEO stock option pay and R&D investments is proven to be positive, significant and robust.

Therefore, it can be assumed that this relation is existent and evident. The relation between CEO stock option pay and stock price volatility is positive, significant and robust. This does not count for the average CEO stock option compensation of the prior three years in relation to the actual year of stock price volatility. This relation appears to be non-significant and thereby reduces the actual robustness of the relation to a certain degree. This section discusses the interpretation of the mentioned results in light of the literature examined.

From an agency theory perspective, both parties to the relation are utility maximizers. To mitigate the differing interests of the principal and the agent, stock options are granted to CEOs by the remuneration committee on behalf of the Board of Directors and ultimately, the

shareholders. Depending on the organization’s attitude towards risk, the principal in the contract will try to allocate agency costs in such a way that the behavior and organizational outcome caused by the agent is controlled to a certain extent. Increased CEO stock option pay, subject to the incentives part of agency costs, will increase organizational risk-taking behavior and thus is suitable for less risk-averse characterized organizations. Risk-averse organizations, in contrast, should not engage in the extensive use of stock option pay as part of CEO compensation.

(33)

33 behavior and/or outcome of the agent, may indeed (Gottschalg & Zollo, 2007) gain a competitive advantage by means of the optimal level of CEO stock option pay. Namely, if the stock option incentive scheme is optimal, interest alignment is enhanced and agency costs of monitoring the agent by the principal are minimized. Minimized agency costs may implicate an improved competitive position of the firm, reflected by an enhanced financial performance.

Managerial self-interest has been proven to be an extant phenomenon (Bouwens & Kroos, 2010). The results forthcoming from the data analysis section of this thesis support the concept of managerial self-interest as CEOs tend to take more risk for the company as a whole, if there is a greater part of their personal potential wealth tied to the stock price. A proclaimed factor (Serfling, 2014) for risk-taking behavior, CEO age, was added to the statistical analysis as control variable because it is expected to influence the two indicators of risk-taking behavior. The results generated by the analysis do not support the findings by Serfling (2014). The regressions performed in the data analysis section show that CEO age is positively related with R&D investments, indicating that older CEOs tend to take more risks. The results of CEO age and stock price volatility do not present a significant relationship. The conflicting results between Serfling (2014) and this study may be caused by the sample in use, covering different companies and different timeframes. Serfling (2014) uses an extensive sample with 4,493 unique CEOs over a timeframe covering the years 1992 to 2010. However, the author excludes the firms from the utilities industry and the financial industry. This may damage the generalizability of the results to some extent. On the contrary, the results following from the regression analysis in this thesis are based on a rather small sample and do not pass the robustness check concerning the relation between CEO age and R&D expenses. Adding the fact that there are no significant results in the relation between CEO age and stock price volatility, it may be concluded that the evidence to support a positive relation between CEO age and organizational risk-taking is very weak. Although this is an interesting topic for a more extensive future research, the weak results are not of crucial importance for the research question guiding this thesis.

Shareholders, who are to be found value maximizers, often have the desire that CEOs undertake more risks than they naturally would be intended to do. As expected, CEO stock option pay, as initiated on behalf of the shareholders, increase the two indicators of

organizational risk-taking behavior. This too, proves the functioning of stock options as an interest alignment tool. However, shareholders and the Board of Directors should be careful with

(34)

34 the granting of stock options to CEOs as organizational risk-taking increases the potential for abnormal returns. As Sanders & Hambrick (2007) proclaim with statistical evidence in their research, the heavy use of CEO stock options brings about more big losses than big gains. This implies that shareholders, who are interested with the total returns of their investments and not only the stock price, should perform a careful analysis of the CEO remuneration policy

concerning the potential investment candidates.

Regarding CEO remuneration policies, the established relation between CEO stock option pay and risk-taking may have serious implications. Firstly, apart from the organizational culture and actual attitude towards risk, shareholder meetings should clarify the interests of the shareholders and their view on organizational risk. The Board of Directors, consequently, should be held responsible for the representation of the shareholders and the provision of strong

corporate governance. Remuneration committees should be in place, an external auditor should be independent (as enforced by SOX), and material weaknesses should be reported. If these peripheral issues are covered by the corporate governance, and earnings management is ruled out to a great extent, CEO remuneration policies can definitely reap the benefits desired by the shareholders and the company as a whole. There is no one best way of remuneration, I would advise on taking a contingency approach. In deciding on the allocation of different compensation components, determining the expected total benefits associated with the job, and defining the relation between pay and performance, the remuneration committee should keep in mind the characteristics of the CEO who is desired. This implies that it is possible that firms carry out a policy that attract CEOs who are by nature less risk-averse. If this appears not to be the case, the additional granting of stock options can still generate the increased risk-taking behavior as desired. Note that companies in need for severe innovation may also use CEO stock option pay to generate internal R&D investments. Remuneration committees should keep in mind that the danger underlying the heavy use of stock option compensation is the fact that there is no economic downside of the stock options for the CEO in question because the CEO can choose not to exercise the option at vesting date. This entails that CEOs can pursue in taking on very risky projects of which the chance of success is very small. In addition, having in mind that disclosure of information about R&D investment projects can increase the stock price, CEOs may take on projects for which the internal rate of return (IRR) does not exceed the cost of capital (CoC). If so, a smaller IRR than CoC indicates that the project will not yield additional

CEO stock option pay and organizational risk-taking behavior in the post-sox era

MSc Thesis: