MSc Accountancy & Control, variant Control
Faculty of Economic and Business, University of Amsterdam
The effect of the Sarbanes-Oxley Act on changes in the composition of the compensation of CEOs of utility firms; and how this affects their investment
decisions.
Student name: Arie Brouwer Student number: 10274006
Date and place: Amsterdam, July 20th, 2015
Program: MSc Accountancy & Control Specialization: Control
Institution: University of Amsterdam, Faculty of Economics and Business
Word count: 10757
First supervisor: Dr. Ir. S. B. H. Morssinkhof Second supervisor: Dr. Ir. S. P. van Triest
1
Statement of Originality
This document is written by student Arie Brouwer 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.
2
Abstract
This thesis studies the effect the passage of SOX has on the composition of the compensation of CEOs of utility firms and how this affects their risk-taking investment decisions. Our sample consist of utility firms (SIC between 6000 and 6999) in the years 1992-2004. We find a negative relation at a significant level of 1% (t = -3.18) between the passage of SOX and the ratio of incentive to guaranteed compensation. This result indicates that firms responded with a lower ratio of incentive to guaranteed compensation for executives after the passage of SOX and supports our hypothesis that environmental changes on the composition of compensation induces higher guaranteed income. It could however be the case that this effect is not solely to the passage of SOX, but is also influenced by the bad publicity for high earning CEOs,
because of recent corporate scandals.
We find an insignificant relation between the passage of SOX and the amount of risky investments. This result indicates that the risky investments of firms did not diminish after the passage of SOX. We suspect this to have something to do with the fact that utility firms are a low-technology sector. Because utility firms are less based on research and long-term horizon expenditures, it could be the case that environmental changes have less effect on capital expenditures. We interpret the results we found in this research as evidence of some of the potential costs of this new act.
3
Table of Contents
Statement of Originality ... 1 Abstract ... 2 1. Introduction ... 4 1.1 Background ... 4 1.2 Research Question ... 81.3 Motivation of this Study ... 8
1.4 Structure ... 10
2. Literature Review and Hypotheses ... 11
2.1 Agency Problems ... 11
2.2 Executive Compensation ... 12
2.3 The Effect of Compensation and Environmental Changes on Investment Strategies .... 15
2.4 Hypotheses ... 17 3. Research methodology ... 18 3.1 Sample Construction ... 18 3.2 Research Method ... 19 3.3 Data Sample ... 22 4. Results ... 24
4.1 Executives’ incentive to salary compensation ratio ... 24
4.2 Executives’ risk-taking investment decisions ... 26
5. Conclusion ... 30
6. References ... 32
4
1. Introduction
This section contains an extensive explanation of how the passage of the Sarbanes-Oxley Act came about, and its specific regulations which indirectly caused a change in the composition of compensation plans for industrial companies. After this background information the introduction continues with the research question and the motivation for this study.
1.1 Background
In March 2001, the longest boom in U.S. history ended. During such a period shareholders will anticipate high growth and this will be built into performance expectations, like: budgets, earnings, stock prices, debt commitments and revenue forecasts. Under these circumstances, managers experience strong financial pressures to deliver growth in earnings and stock market performance, to maintain their bonuses or current position (Ball, 2009).
There is the possibility that during this period of high growth corporate monitors (like analysts, boards, rating agencies and the press) regard high growth as a given factor. This will increase the risk of slow reactions when this high growth is not ‘normal’ anymore. When this boom ‘busts’ and growth suddenly falls, managers will be unable to meet the expectations to maintain their bonuses and/or current position. This will give some managers a drive to disguise their lower performance by using unaccepted accounting methods and reporting false transactions (Ball, 2009).
These kinds of ‘financial frauds’ appear to have three reasons. The first one is the inability to meet expected performance. There are less reasons for fraud when a manager is able to meet its targets. Second, the existence of personal costs when failing to deliver these expected performance. The prospect of losing your current position or bonus package causes some managers to disguise their lower performance. Third, the idea that real performance will improve soon. Else deception must be repeated to avoid detection, which is not sustainable (Ball, 2009).
External auditors could have a big influence on these kinds of unaccepted accounting methods and unethical accounting. For external auditors, there is however a conflict of interest in performing non-audit work for clients. The auditor may be influenced by having a direct or indirect financial interest in the client. Fraud can happen when an auditor is
performing significant consulting engagements for a client. The auditor may be reluctant to insist on accounting adjustments because of the fear of losing the client to another audit firm. By giving clients favorable audit treatment, they can retain lucrative consulting engagements.
5 The problem here is that auditors supply a certification that is used by every party of interest, and it is left to the audit firm itself to owe a standard of care. This gives the possibility that audit fees themselves may cause a drift toward fraud. (Claypool, Tackett & Wolf, 2004).
This problem in auditing showed up in the case of Enron1 v. Arthur Andersen2. Enron was a successful energy trading company, with revenues that exceeded $101 billion in 2000, making it the seventh-largest company in the US. However, to sustain its growth and keep the stock prices at a high level, Enron falsified statements and hid losses through complex
accounting methods. These deceptive financial maneuvers that had be repeated to avoid detection, were exposed in the end of 2001. Enron’s stock price dropped dramatically and the firm went bankrupt in December 2001. As the SEC3 began investigating the company’s records, Enron’s auditor Arthur Andersen shredded relevant documents. The audit firm collapsed after being indicted by the Justice Department. The case of Enron v. Arthur Andersen was however just the first of many. Now that the public and the SEC had an idea where to look, many other firms reported hidden losses before being exposed by an external party. WorldCom, the nation's number two long-distance phone company, reported to have improperly booked $3.8 billion in expenses (Ailon, 2011).
The regulatory and political response to the vast amount of scandals that were exposed during 2001-2002 consisted of two categories. Criminal proceedings against companies and individuals, like the criminal conviction in 2006 of Enron’s former CEOs, Jeffrey Skilling and Kenneth Lay, on counts of fraud and conspiracy. The other political response were legislative proceedings, meant to reduce the incidence of negligence and fraud in the future (Ailon, 2011).
This legislative response to the accounting scandals was the Sarbanes-Oxley Act, signed into law on July 30, 2002. The Sarbanes-Oxley Act or ‘SOX’ provides the most extensive regulation of the securities markets since the Securities Act of 19334.
The provisions of the Sarbanes-Oxley Act include(Ball, 2009):
1 Enron is a an energy trading firm that bought and sold gas and electricity, and ventured abroad to build power plants and utilities. Enron eventually became a dot-com, trading commodities online.
2 Until 2001, Arthur Andersen was one of the biggest auditor firms in the world. 3 Securities and Exchange Commission.
4 This act created amongst other things the Securities and Exchange Commission and was also a response to the political crisis of the Great Crash of 1929.
6 - Mandatory certification of financial statements and internal controls by CEOs and
CFOs, with criminal penalties for irresponsible certification;
- An independent audit committee of the board must be established, which is responsible for overseeing and the hiring of auditors;
- Misleading or influencing auditors to provoke fraud is prohibited;
- Auditors are prohibited from non-audit work which may compromise their independence, for example Bookkeeping;
- A mandatory rotation of audit partners;
- A compulsory disclosure of codes of ethics, off-balance sheet transactions and internal controls;
- A prohibition of publishing reports on companies by security analysts with conflicts of interest;
- The creation of the PCAOB5 to oversee the audit industry
The passage of the Sarbanes-Oxley Act triggered a fierce debate about its net effects on the US economy. This legislation started as a response to corporate scandals, but critics are concerned about high compliance costs and the unintended consequences of SOX regulations. Therefore, whether SOX is successful depends on the trade-off between the benefits and costs of the regulation (Kang, Liu & Qi, 2010). From a regulatory point of view, the downside of SOX were its short-term compliance costs, in particular for smaller companies (Ball, 2009). Zhang (2005) discusses the private costs and benefits of SOX, using empirical tests. By calculating the cumulative abnormal returns around the legislative events6 she examines the short-term changes of the market index. The results suggested that the passage of SOX had severe economic consequences and that there was a loss of $1.4 trillion in the total market value during the most significant events of rulemaking (Zhang, 2005).
Agency theory suggests that firms can align the interests of managers and shareholders through bonus contracts (Jensen & Meckling, 1973). The firm frequently designs bonus contracts based on earnings to encourage effort by the manager. The problem is that managers may control both effort and the reporting of earnings. Since these contracts use mostly
5 The PCAOB is a regulator of the audit industry. It is empowered to enforce legislations that may affect the whole audit industry, including auditing procedures, practices and standards.
6 Amongst other things, the passing of the Sarbanes’ bill by the Senate Banking Committee and the request of the SEC that CEOs and CFO’s of the largest 1000 firms would certify their financial reports.
7 earnings as a way to measure a manager’s effort, a contract that encourages effort may also encourage earnings management. This way, firms had to make a tradeoff between the
incentive for effort and the possibility that managers will not report truthfully (Carter, Lynch & Zechman, 2009).
In 2002 however, with the passage of SOX, there were greater penalties for this kind of managerial misconduct. Because of all these new regulations and broader financial reporting responsibilities, CEOs faced a higher risk from misstatements of financial
information, by themselves or their employees. The tradeoff between the incentive for effort and the possibility that managers will not report truthfully shifted (Cohen, Dey & Lys, 2004).
Cohen, Dey & Lys (2004) hypothesize that firms were more likely to alter bonus incentives to compensate their CEOs for these added risks. Firms could respond in different ways, for example by increasing compensation levels to maintain the CEOs original utility level. However, there has been a lot of criticism of the high levels of CEO compensation. The recent accounting scandals have also focused attention on this issue. Another way could be an increase in salary and bonus compensation, and decreases in option-based compensation. An increase in fixed salary and bonus compensation results in less risk than option-based
compensation, which compensates for the added risks, because of the passage of SOX. Less incentive based compensation could however diminish the alignment between shareholders and managers (Cohen, Dey & Lys, 2004).
By looking at a sample selected from industrial companies, excluding utilities,
financial, and transportation firms from 1992 to 2006, Cohen, Dey & Lys (2004) document an increase in equity incentives of CEOs over time, which declined significantly after SOX. This trend is likely to be due to the decline in option-based compensation, which was an essential part of the equity incentives. Cash compensation decreased over time, but increased
significantly after SOX. They also found that the proportion of incentive-based compensation to fixed salary and the pay for performance sensitivities declined after SOX. This shift was likely established to shield executives from some of the risks imposed by SOX.
This new regulatory environment and the changes in the compensation plans and incentives of CEOs may alter their investment decisions. It may give management the drive to deviate from value-maximizing actions. The significant changes in reporting responsibilities of management and the increased risk because of this, could reduce risk-taking activities by managers. Incentives to undertake risks are lessened when managers face more severe penalties for bad results (Cohen, Dey & Lys, 2004). Cohen, Dey & Lys (2004) document an increase in risky investments over time, which declined significantly after the passage of
8 SOX. This evidence could support the hypothesis that the passage of SOX increased risk-aversion on the part of executives, which caused a decrease in spending on more riskier projects.
Kang, Liu & Qi(2010) do research toward a similar topic and use the Euler equation framework to determine the effects of the Sarbanes-Oxley Act on firms’ corporate investment decisions. By using a sample of industrial firms, from Standard & Poor’s Compustat Annual database, covering the period 1998 to 2005, they find robust evidence that the discount rates managers apply rise significantly after SOX. They however omit all financial and utility firms, because their model is inappropriate for them (Kang, Liu & Qi, 2010).
1.2 Research Question
There has been a lot of research into the consequences of the Sarbanes-Oxley Act on total market value, bonus incentives and compensation, abnormal stock return and earnings management. This research will however focus on the implications of changes of compensation plans, on the risk-taking investment decision making by management.
We examine trends in the levels and composition of fixed and incentive-based
compensation and changes in the ratio of incentive to fixed compensation after the passage of SOX. Next we look at managers’ investment policies after the passage of SOX and whether compensation and risky investment are related to future return on assets after the passage of SOX.
Cohen, Dey & Lys (2004) are the only ones who have done research on this topic, but they totally ignore the effects of SOX on the composition of compensation plans in the utility sector and the change in investment decisions. This makes the research question for this study:
What is the effect of the Sarbanes-Oxley Act, on changes in the composition of the
compensation of CEOs from utility firms; and how does this affect their investment decisions?
1.3 Motivation of this Study
The purpose of this study is to investigate changes in the composition of the compensation of CEOs and how these changes affect the investment decisions in utility firms after the passage of SOX. If the composition of compensation plans for CEOs in financial firms were optimal before the passage of SOX, changes could result in deviations from value-maximizing actions.
9 This study is based on research done by Cohen, Dey & Lys (2004). They hypothesize that firms are more likely to alter bonus incentives to compensate their CEOs for the added risks of SOX. They also hypothesize that these changes in compensation plans have a direct influence on their risk-taking investment decisions. When compensation structures consist of more short-term rewards, managers decisions become less aligned with the companies, which will influence their investment decisions. They find with a significance level of 5% (t = -2,09) a decrease in the ratio of incentive compensation (value of options + bonus compensation) to fixed compensation (guaranteed salary) after the implementation of SOX. They also find with a significance level of 5% (t = -2.12) a decrease in risk-taking investments (R&D expenses as a proportion of sales) after the passage of SOX.
According to Murphy (1999) there are significant differences in the total amount and composition of compensation for CEOs between the mining and manufacturing firms, financial services, utilities and other industries. The average pay increase of a CEO of an utility firm in the S&P500 in 1996 consist of 18% guaranteed income, 49% options and restricted stock and 33% stock. Median payment increases of CEOs of other industries consist of 3% guaranteed income, 37% options and restricted stock and 38% stock. There are also huge differences in stock ownership7 and the composition of total compensation8 between utility and other industries.
These are significant differences in the composition of compensation plans and how they change when the total compensation changes. If the optimal levels of compensation were different before SOX, then the effects of SOX and the increased risk for managers could also have different implications for the utility industry. Cohen, Dey & Lys (2004) however, do not include any utility firms in their sample. These observations could give other results than those from Cohen, Dey & Lys (2004), when researching this specific industry.
For example, McConnell and Muscarella (1985) did research on the market reaction due to capital expenditures by utility and industrial companies. They found that
announcements of increases or decreases in capital expenditures9 had a significant effect on stock returns for industrial companies. In the utility sector however, announcements of capital expenditure did not have any significant effect on stock returns. Because utility firms are less
7 Utility in 1996: 0,0225% ; Other industries in 1996: 0,13%
8 Utility in 1996: 40% salary, 20% bonus, 17% options and 23% other ; Other industries in 1996: 28% salary, 19% bonus, 36% options and 17% other
10 based on research and long-term horizon expenditures, it could be the case that the passage of SOX has a smaller effect on utility firms in comparison with the industrial sector.
1.4 Structure
The remainder of this paper will be organized as follows. Section 2 reviews the literature. This part will be dedicated to Agency Theory, how the executive compensation is determined and how investments may be influenced by this. In addition this section will contain the hypotheses for this study. Section 3 examines the methodology for this study. It will consist of the research method we want to employ, how we retrieve our data sample and the main theoretical constructs. Section 4 presents the results of the empirical tests of our hypotheses. Section 5 summarizes, concludes and proposes several limitations and suggestions for future studies.
11
2. Literature Review and Hypotheses
In this section an overview will be given of the composition of executive compensation and its effect on investment decisions. This section will start with an explanation of agency problems and the reasons for different kinds of compensation. After this the composition and growth of executive compensation will be discussed. We will look at changes of the
compensation packages when influenced by environmental changes. Different variables will be mentioned which have an influence on the composition and size of compensation plans and risk-taking investment decisions. Also, the main empirical results of important studies about the composition and total compensation of CEOs will be reviewed. Finally, we state our hypotheses.
2.1 Agency Problems
Agency problems arise when there is a separation of ownership and control. This is because contracts between the agent and the principal cannot be written and enforced without costs. The costs of monitoring, structuring and bonding of the contracts for the agent (with
conflicting interests), fall under agency costs. It is important to control these agency problems, when the manager who is in control does not share major wealth effects of these decisions in question. Without an effective control of these agency problems, managers are more likely to follow their own interests and deviate from the interest of the principal(Fama & Jensen, 1983).
To limit these divergences of the agent, the principal can either monitor the agent, with the extra costs that come with it, or establish incentives which motivates the agent to align their goals and diminish the conflicts of interests between the principal and the agent. In some cases the principal will pay the agent to expend his resources in the firm (bonding costs) to make sure that the agent will not take actions which would harm the principal. But without making any costs, it is almost impossible to let the agent make optimal decisions from the principal’s perspective. There will always be a difference between the decisions made by the agent and the ones which would maximize the principal’s welfare (Jensen & Meckling, 1973).
The compensation plans of executives can be seen as one of those incentives to align the goals between the agent and principal. Compensation practices are often seen as
organizational responses to agency problems, where the owners want to extract effort from the agents, when dealing with information asymmetries and conflicts of interest. Agency theory suggests that managers are risk-averse, and that providing effort has a cost. Managers
12 put the amount of effort in their work to keep their job, but not more. Firms can align the interests of managers and shareholders in a cheap way through their compensation plans, with bonding costs, like stocks and options. This way there is a constant search for an optimal tradeoff between efficient risk sharing and efficient effort of the agent, while making a contract (Ortín‐ángel & Salas‐fumás, 1998).
2.2 Executive Compensation
The structure of CEO compensation is designed in such a way that executive decision making is directed toward maximizing firm performance. There was however a rapid growth in executive pay, which started in the mid-1970s and continued until the beginning of the 21th century. The increase in compensation was at its high point during the 1990s, with annual growth rates of more than 10%. In less than ten years, the median level of CEO compensation climbed from $2.3 million in 1992 to $7.2 million in 2001, and remained stable in the next few years10 (Frydman & Jenter, 2010).
According to Frydman & Jenter (2010), the five most basic components of CEO compensation are: salary, restricted options, stock grants, annual bonus and payouts from long-term incentive plans. The importance of each component changed drastically over time. From 1936-1950, CEO compensation consisted mostly of salaries and annual bonuses and in the 1960s a focus came on long-term rewards which were paid out over several years. In the 1980s, there was a comprehensive increase in stock option compensation, with the purpose to increase incentives to increase shareholder value11. The amount of executive option
compensation increased from 20% in 1992 to 49% in 2000. So the great increase in total CEO compensation was mainly driven by this increase in stock options (Frydman & Jenter, 2010).
Although the increase in total CEO and stock option compensation were occurring in the entire industry, there were still major differences in CEO compensation between the different industries. When looking at the trend of stock ownership for S&P 500 CEOs between 1987 and 1996, there are significant differences between the utility industry and other industries. Murphy (1999) did research on the median percentage of stock ownership by CEOs in different industries. He found a stock ownership of 0,045% ($0,9 million) in utility firms in 1987 which dropped to 0,0225% ($1.5 million) in 1996. In other industries the CEOs hold a median percentage of stock ownership of 0,35% ($4,8 million) in 1987 and 0,13% ($7
10 The research of this paper is based on the years 1992-2004, which is the reason we do not expand further on later years.
11 This was made possible by a tax reform which permitted option payoffs to be taxed at a much lower rate than the taxable rate of labor income.
13 million) in 1996. This large difference in stock compensation would suggest a smaller
incentive for alignment in the utility industry. More of these differences can be seen in the level of total CEO compensation12 and the composition of CEO compensation13 (Murphy, 1999).
The significant rise of CEO compensation over the past 40 years is explained by many different theories. The increase in CEO compensation could be a result of the ability of an executive to set his own compensation and extract rents from the managed firm. Rent extractions is possible because of weak corporate governance where CEOs are allowed and have the power to set their own pay. Other theories consist of CEO effort being a scarce managerial talent for which large companies want to pay an excessive amount (Frydman & Jenter, 2010). Managerial talent has become a scarce resource, because of more complicated and bigger organizational structures. Larger firms are more complicated and need better governance and managerial skills. This higher need for managerial talent comes at a price. This makes the size of a company an important variable for deciding the compensation of executives. Incentive compensation is used in particularly to align the interests between shareholders and management in these big organizational structures (Fredric & Taylor, 2004)
Other variables which influence the composition of compensation can be seen by comparing Murphy’s (1999) summary of CEOs pay-performance sensitivities and the level and composition of total CEO compensation. The average pay increase of a CEO of an utility firm in the S&P500 in 1996 consist of 18% guaranteed income, 49% options and restricted stock and 33% stock, while his total compensation consists of 60% guaranteed income, 17% options and 23% stocks and other income. This means that the composition of compensation changes over time and that the amount and composition of compensation for CEOs depends on the amount and composition of compensation the CEO currently gets (Murphy, 1999).
A final market-based explanation for the growth in CEO compensation is improved monitoring and stricter corporate governance (Frydman & Jenter, 2010). Hermalin (2005) shows that an increase in monitoring, which has a negative effect on the job security of a CEO, leads to an optimal response by the firm to increase the level of CEO compensation.
Basu, Lal, Srinivasan & Staelin (1985) hypothesize about environmental changes and its effects on compensation. When environmental uncertainties increase, there are greater
12 Utilities: $1,133 million in 1992 and $1.523 million in 1996 Other industries: 2.204 million in 1992 and $3.624 million in 1996
13 CEOs in the utility industry are paid perceptual considerably more in salary than in options and also perceptual more in bonuses, starting from 1995, in comparison to other industries.
14 chances for a high or low outcome, given a certain level of effort. A risk averse agent will in this case demand higher compensation for the same amount of effort, which results in a higher marginal cost per unit of effort for the firm, with no higher corresponding increase in
expected revenues. In this case the principal has no benefit from motivating the agent beyond a certain point of effort, because the highest profit from an employee is gained when marginal costs are the same as marginal revenue. This way the costs for a higher effort outweigh the extra revenues gained. It would be optimal for the firm to obtain a lower level of effort from the agent (which means lower compensation), with expected reduced revenues. The agent however still wants a higher guaranteed compensation because of the increased environmental uncertainties. This is why the firm offers the agent an increase in the proportion of salary (guaranteed income), while reducing the total compensation (Basu, Lal, Srinivasan & Staelin, 1985).
This theory was tested by Campbell, Umanath & Ray (1993), with a laboratory experiment. They use the hypothesis that an increase in environmental uncertainty will cause a decrease in expected compensation and at the same time an increase in the guaranteed compensation to incentive compensation ratio. As sample they use 79 first-year MBA students. Each student played the role of a principal and was tasked to award compensation contracts to multiple agents working for them in different environmental situations.
By using repeated measure analysis of variance, they document that when the
employee/agent is well informed, total compensation may be reduced when there is perceived environmental uncertainty. To compensate the agent, the firm should increase the proportion of base salary (guaranteed income) in total compensation. However, when the agent is not fully informed (which is a more practical scenario), the results suggests either an increase in the guaranteed compensation to incentive compensation ratio or an increase in the total compensation. These findings partially contradict the hypothesis and the theory by Basu, Lal, Srinivasan & Staelin (1985). This contradiction may be caused by a lack of control in the amount of knowledge both the principal and the agent have (Campbell, Umanath & Ray, 1993).
15
2.3 The Effect of Compensation and Environmental Changes on Investment
Strategies
Although investment decisions and compensation policies are endogenous, agency theory suggest that incentive compensation aligns the interests of managers and investors, which means that investment decisions will also be influenced by the composition of CEO compensation (Jensen & Meckling, 1973).
Ryan & Wiggins (2002) address this issue by looking at the influence of incentive compensation on R&D. R&D investment is characterized by a high degree of uncertainty and a long-term horizon, which makes it a measure for risk-taking investment decisions. When managers are compensated for short-term cash flows, there will be little investment in R&D (the long-term horizon). Their results suggest that R&D is positively related to equity-based compensation, because equity-based compensation focuses on a long-term outlook and provides incentive alignment. These results would indicate that risky investments/R&D expenses would drop when CEOs are offered higher guaranteed compensation and lower incentive compensation. However stock options and restricted stock have opposite effects on R&D. Options encourage risky investment, because nonlinear payoffs minimize downside risk. Linear payoff of restricted stock on the other hand, discourage risky investment (Ryan & Wiggins, 2002).
Risk-taking investment decisions are influenced by more than just the ratio of incentive compensation to fixed compensation. Fisher & Temin (1973) explain with the Schumpeterian hypothesis that size is significantly positively related to R&D expenses. Because large firms tend to have more diversified activities than smaller firms, they have a smaller risk of producing knowledge they cannot use. When firms are engaged in fewer activities they have less to gain from R&D. Large firms also enjoy economies of scale in the financial market, because they are able to borrow more money at a lower cost. Frenkel & Shefer (2005) however, find that there is only a relation between investment in R&D and firm size in high-tech firms, but they find for other firms only insignificant observations.
Another way the amount of more risky investments can be measured is through capital expenditures, since capital expenditures are both characterized by a high degree of uncertainty and a long-term horizon. Eli, Gilad & Yanling (2007) estimate the association of investments in R&D and in physical assets when looking at operating income. Their results support a fundamental difference between investment information about capital expenditures and R&D. They find that, on average, capital expenditures contribute less to operating income than R&D. This is however only noticeable in industries which are R&D-intensive. In industries
16 that are based more on low-technology, like the utility sector, there are no real statistical differences between R&D and capital expenditures and their relation towards risk-taking investment decisions (Charoenwong & Chung, 1998).
The age of the CEO is also an important factor in the amount of risk-taking
investments decisions made. Dechow & Sloan (1991) show that CEOs nearing retirement, reduce risk-taking investments decisions. This is presumably the case to increase reported earnings in their final years. They also state that older CEOs may have fewer incentives to undertake risky investments.
Dechow & Sloan (1991) also find significant evidence for the lame-duck hypothesis. This hypothesis states that CEOs delay new investment decisions when a CEO change is imminent, so that the successor still has a certain kind of flexibility. This hypothesis predicts that important investment decisions take place early in the tenure of the CEO and that CEOs become ‘lame ducks’ when they prepare to hand the firm over to their successor (Dechow & Sloan, 1991).
Changes in the regulatory environment and the composition of compensation plans of CEOs are likely to have an influence on their investment decisions. For example, managers who bear great residual risk because of these regulatory changes, want to reduce the residual uncertainty of their investments by making less risky investments. This way environmental changes may give management the incentive to deviate from value-maximizing actions for the firm. Significant changes in reporting responsibilities of management and the increased risk because of this, could reduce risk-taking activities by managers. Incentives to undertake risks are lessened when managers face more severe penalties for bad results The implementation of the Sarbanes-Oxley Act results in such changes (Cohen, Dey & Lys, 2004).
Kang, Liu & Qi (2010) did research into the investment decisions of managers after SOX. They however omit all financial and utility firms, because their model is inappropriate for them. By comparing the discount rates used by managers of 1072 firms in the U.S. and the U.K. before and after SOX, they find robust evidence that the discount rates managers apply rise significantly in the U.S. after the passage of SOX (a great environmental change). The estimated coefficient on the U.S. is significantly negative (t = -0,983), which suggests that U.S. firm managers perceive a higher discount rate than U.K. managers over the sample period. The discount rates for managers in the U.K. remain virtually unchanged over the whole sample period. Higher discount rates for all investments result in a lower amount of investments which seem profitable, with less total investment as a consequence (Kang, Liu & Qi, 2010).
17
2.4 Hypotheses
This research wants to examine trends in the ratio of fixed and incentive-based compensation after the passage of SOX. Next, we look at managers’ investment policies after the passage of SOX and whether the risk-taking investment decisions have changed, because of a change in compensation and the passage of SOX.
When environmental uncertainties increase, a risk averse agent will demand a higher compensation for the same amount of effort. In this case however, the principal has no benefit from motivating the agent beyond a certain point of effort and it would be optimal for the firm to gain a lower level of effort from the agent, which means lower compensation (Basu, Lal, Srinivasan & Staelin, 1985). SOX leads to these environmental uncertainties for CEOs, by imposing new regulations and broader financial reporting responsibilities and greater penalties for managerial misconduct.
The agent still wants higher guaranteed compensation when there are increased environmental uncertainties. This is why the firm offers the agent an increase in the
proportion of salary (guaranteed income), while reducing the total compensation (Basu, Lal, Srinivasan & Staelin, 1985). So our first hypothesis is:
H1: There is a significant decrease in the ratio of incentive compensation (value of options +
bonus compensation) to fixed compensation (guaranteed salary) in utility firms after the passage of SOX when compared to the period prior to SOX.
Next, we look at managers’ investment policies after the passage of SOX and whether compensation and risky investment are related to future return on assets after the passage of SOX.
Changes in the regulatory environment and the composition of compensation plans of CEOs are likely to have an influence on their investment decisions. Managers who bear greater residual risk, because of regulatory changes like SOX, want to reduce the residual uncertainty of their investments by making less risky investments. This makes our second hypothesis:
H2: There is a decrease in risk-taking investments as a proportion of sales in utility firms
18
3. Research methodology
This section provides a clear and elaborate description of the sample we use in this study and the way we retrieve the data. We also explain which research method we want to employ and the way we want to empirically test our main theoretical constructs to give an answer to our hypotheses.
3.1 Sample Construction
This study only looks at firms that have primary Standard Industrial Classification (SIC) between 6,000 and 6,999. These are all firms in the utility industry. The sample for this study will be selected from the S&P 1500, because picking firms from the S&P 500 for this specific research may lead to a very small sample. The S&P 1500 covers approximately 90%14 of the U.S. market capitalization and replicates the performance of the U.S. equity market against a representative universe of tradable stocks, which would suffice for this study. We only look at the U.S. market, because this is where the Sarbanes-Oxley act was enacted.
All the necessary data can be obtained through multiple databases. This gives the research method of this study an archival purpose. We use WRDS to gain access to the database COMPUSTAT and EXECUCOMP. COMPUSTAT will provide us with all firm specific information needed for our model and we use EXECUCOMP to gain all CEO related information. Merging the COMPUSTAT and EXECUCOMP databases will result in our sample of firms and firm-year observations.
The data for this study are annual, covering the period from 1992 to 2004. We choose 1992 because EXECUCOMP does not have data prior to this year and this way we cover 10 years before the passage of SOX to get a large enough sample to find significant changes in comparison to the sample of 3 years during and after the passage of SOX. 2004 has been chosen as the final year, so that we have a large enough sample to look for differences before and after the passage of SOX, but other environmental economic changes like the financial crisis in 2008 would not play a role in our results.
We apply several data filtering rules to obtain our final sample. We drop sample firms with missing data for any of the included endogenous and exogenous variables, or find an alternative method to obtain or replace the missing values. This will give the reduced and final sample size. A problem of using EXECUCOMP to get the data for CEO compensation is
19 that the sample will likely be dominated by larger firms, because EXECUCOMP does not contain CEO information of smaller firms.
3.2 Research Method
To empirically test our research hypothesis and find out what the effect of the Sarbanes-Oxley Act was, on changes in the composition of the compensation of CEOs of utility firms; and how this affects their risk-taking investment decisions, we want to use an adjusted two stage least squares (2SLS) model by Cohen, Dey & Lys (2004).
Although investment decisions and compensation policies are endogenous, agency theory suggest that incentive compensation aligns the interests of managers and investors, which means that investment decisions will also be influenced by the composition of CEO compensation (Jensen & Meckling, 1973). This makes studying the ratio of incentive compensation to fixed compensation and defined as the proxy for risky investments of
executives separately, redundant and with inferences. The equations for the 2SLS regressions are the following:
(1)𝑅𝐴𝑇𝐼𝑂𝑗𝑡=∝0+∝1× 𝑆𝐼𝑍𝐸𝑗𝑡+∝2× 𝑅_𝐼𝑁𝑉𝑗𝑡+∝3× 𝐴𝑂𝐶𝑗𝑡+∝4× 𝐶𝐹𝑂𝑗𝑡+∝5× 𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡 +∝6× 𝑇𝐼𝑀𝐸 +∝7× 𝑆𝑂𝑋 +∝8× 𝑅_𝐼𝑁𝑉𝑗𝑡× 𝑆𝑂𝑋
(2)𝑅_𝐼𝑁𝑉𝑗𝑡=∝0+∝1× 𝑆𝐼𝑍𝐸𝑗𝑡+∝2× 𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡+∝3× 𝐶𝐹𝑂𝑗𝑡+∝4× 𝐴𝐺𝐸𝑗𝑡+∝5 × 𝑇𝐸𝑁𝑈𝑅𝐸𝑗𝑡+∝6× 𝑅𝐴𝑇𝐼𝑂𝑗𝑡+∝7× 𝑇𝐼𝑀𝐸 +∝8× 𝑆𝑂𝑋 +∝9× 𝑅𝐴𝑇𝐼𝑂𝑗𝑡 × 𝑆𝑂𝑋 +∝10× 𝐴𝐺𝐸𝑗𝑡× 𝑇𝐸𝑁𝑈𝑅𝐸𝑗𝑡
𝑅𝐴𝑇𝐼𝑂𝑗𝑡 is defined as is the ratio of incentive compensation (value option grants15 plus bonus compensation) to fixed compensation (guaranteed salary). The values of these variables can be obtained through EXECUCOMP’s BONUS, SALARY and
OPTION_AWARDS_BLK_VALUE compensation data. Salary must be higher than 0 to be included in this study, to avoid errors in our calculations.
𝑅_𝐼𝑁𝑉𝑗𝑡 is defined as the proxy for risky investments of executives and measured by the R&D or capital expenditures as a proportion of sales. This can be obtained through
15 Calculated by the BlackScholes formula.
20 COMPUSTAT’s SALE and XRD income statement items and the CAPX cash flow item. Sales must be higher than 0 to be included in this study, to avoid errors in our calculations.
𝑇𝐼𝑀𝐸 is defined as the current year minus 1992, which can take a form of 0 (year = 1992) to 12 (year = 2004). This variable TIME has been taken into account to capture the trend over time in the corresponding dependent variables.
jt
SIZE
is defined as the market value of equity at the end of the fiscal-year. This variable is used as a control for firm specific attributes that may influence decisions of managers. Fisher & Temin (1973) explain with the Schumpeterian hypothesis that Size is significantlypositively related to R&D expenses, because large firms have more diversified activities and can enjoy economies of scale. Frydman & Jenter (2010) hypothesize about complicated and bigger organizational structures for which more managerial talent is needed, which has an effect on the compensation of CEOs. The market value of equity will be calculated by multiplying COMPUSTATS annual fiscal closing price (PRCC_F) with the total common shares outstanding (CSHO).
𝐴𝑂𝐶𝑗𝑡 is defined as all other compensation other than bonus, salary and options granted. This variable is used as a control for the fact that the future compensation of incentives depends on the current amount and composition of compensation of the CEO. We expect that companies which use more incentive compensation also use other forms of compensation for the
alignment between shareholders and motivation, because other forms of compensation are another form of incentives (Carter, Lynch & Zechman, 2009). This information can be obtained through EXECUCOMP’s compensation data OTHCOMP.
𝐶𝐹𝑂𝑗𝑡 is defined as the cash flow from operations. This variable is used as a control for the effect of activity on managers’ spending on risky investments and the compensation offered to them. This way we can proxy for the real economic environment and the performance
(Cohen, Dey & Lys, 2004). We obtain the operating activities net cash flow (OANCF) from COMPUSTAT’s cash flow items.
𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡 is defined as the growth in sales for the current year. This variable is used as a control for firm specific attributes that may influence decisions of managers and their
21 compensation structure. Jensen (1986) explains that a growth in sales is positively related to a positive change in compensation. We can calculate the growth in sales with COMPUSTAT’s sales information (SALE) from the current and last year.
𝑆𝑂𝑋 is defined as a dummy variable which takes the value of 1 when the proxy statement is after July 30, 2002, and zero otherwise. Basu, Lal, Srinivasan & Staelin (1985) hypothesize about the effects of environmental changes on the composition of compensation and how more risks induces higher guaranteed income. Kang, Liu & Qi(2010) did research on the investment decisions of managers after SOX and found higher discount rates, which leads to lower total investment.
𝐴𝐺𝐸𝑗𝑡 is defined as the age of the CEO. This variable is used as a control for the fact that older CEOs may have fewer incentives to undertake risky investments (Dechow & Sloan, 1991). The age of the CEO in a given year can be determined by searching in
EXECUCOMP’s executive information for the present age (Present Age) of the CEO and subtracting the difference between 2015 and the year in question.
𝑇𝐸𝑁𝑈𝑅𝐸𝑗𝑡 is defined as the number of months the CEO has been in office on the date of the proxy statement. This variable is used as a control for the fact that CEOs close to retirement may have fewer incentives to undertake risky investments (Dechow & Sloan, 1991). This can be determined by looking at the first day he/she became CEO (Date Became CEO) in
EXECUCOMP and subtract the year in question from it.
The issue of statistical interaction arises in this model, because there are two or more independent variables which affect one another. One of these cumulating variables are our expectations of 𝑆𝑂𝑋 having an effect on 𝑅_𝐼𝑁𝑉𝑗𝑡, because Kang, Liu & Qi (2010) find that the discount rates managers apply rise significantly in the U.S. after SOX, which gives less profitable investment opportunities.
According to our hypotheses 𝑆𝑂𝑋 also has an effect on 𝑅𝐴𝑇𝐼𝑂𝑗𝑡, because increased
environmental uncertainties like SOX motivates the firm to offer the agent an increase in the proportion of guaranteed income, while reducing the total compensation (Basu, Lal,
22 Because CEOs with a higher age have a greater possibility to have a longer tenure, because they worked more years in total, there will be a correlation between the two variables. This statistical interaction will be solved with the extra variable 𝐴𝐺𝐸𝑗𝑡× 𝑇𝐸𝑁𝑈𝑅𝐸𝑗𝑡.
3.3 Data Sample
To start the search for our data sample, we look at all variables the database EXECUCOMP can provide us. We look at the entire database for annual compensation for firms with SIC codes 6000 to 6999, between the years 1992 and 2004. The variables we look at are bonus compensation (BONUS), guaranteed salary (SALARY) 16, value option grants
(OPTION_AWARDS_BLK_VALUE, OTHCOMP), all other compensation (OTHCOMP), the present age of the CEO (Present Age) and the date the CEO came into office (Date Became CEO).
Here we get 19700 observations from 463 different companies. When dropping all firm-years with missing data for any of the included variables we still have 12893
observations from 459 different companies. This also includes omitting every firm year with a salary of 0$, to avoid errors in our calculations. We however did not omit all firm-years for the variable ‘Date Became CEO’ which was meant to be used to calculate the tenure of a CEO. If we would omit all these observations we would lose 2/3th of our observations and only have 4319 observations from 424 firms left. For this reason we decide not to use the variable 𝑇𝐸𝑁𝑈𝑅𝐸𝑗𝑡 in our regression to maintain a larger data sample. Because 𝑇𝐸𝑁𝑈𝑅𝐸𝑗𝑡 and 𝐴𝐺𝐸𝑗𝑡 have a certain correlation we think a larger data sample is more important than maintaining the variable for our regression.
Next we look at all the variables the database COMPUSTAT can provide us. We look at the entire database for North America, Fundamental Annual with SIC codes 6000 to 6999, between the years 1992 and 2004. We look only at the USD currency code. The variables we look at are net sales (SALE), cash flow from operations (OANCF), annual fiscal closing price (PRCC_F), total common shares outstanding (CSHO), research and development expenses (XRD) and capital expenditures (CAPX).
Here we find a total of 46186 observations from 4585 different companies. If we drop all firm-years with missing data of any of the included variables we still have 10728
observations from 2275 different companies. This also includes omitting every firm-year with
23 net sales of 0$, to avoid errors in our calculations. We however did not omit all firm-years with missing data for the variable ‘XRD’ which was meant to be used to calculate the proxy for risky investments of executives. If we would omit all these observations we would lose 90%of our observations and only have 577 observations from 107 firms left. According to Eli, Gilad & Yanling (2007) there are no real statistical differences between R&D and capital expenditures as proxy for risk-taking investment decisions of executives, when the industry is not R&D intensive. Since the industry we look at are utility firms, which is a low
technological sector (Charoenwong & Chung, 1998) and not technological, or pharmaceutical, we omit R&D expenses as a proxy for risk-taking investment decisions and only use capital expenditures, so that we can maintain a larger data sample.
Next we combine the two data samples we currently possess using the ticker symbol of the company and the fiscal year as common values between the two databases. When omitting all firm years and companies the two databases do not have in common we get 7347 observations from 364 different companies.
To calculate the growth of sales per year, we always need the year before for our calculations. Because we do not have the information of all firms before 1992, we need to omit the first year firm year sales, for which the growth in sales cannot be calculated. This gives 6596 observations from 256 different companies.
To get the reduced and final sample size we omit all firm-years for which we do not have any information during the passage of SOX. This is why we omit all companies for which we do not have at least the information of the years 2000, 2001 and 2002. Our final sample size is 4748 observations from 121 different companies.
24
4. Results
Section 4.1 shows the results of the determinants of changes in executives’ incentive to salary compensation ratio. In section 4.2 we present the determinants of executives’ risk-taking investment decisions and end with the results of our instrumental variables (2SLS) regression.
4.1 Executives’ incentive to salary compensation ratio
Table 1 presents the results of the estimating equation (1). Capital expenditures (CAPX) is used as the proxy for risky investments. All variables in the regression are statistically
significantly associated with the incentive to salary compensation ratio, except for the variable 𝑅_𝐼𝑁𝑉𝑗𝑡× 𝑆𝑂𝑋.
The coefficient for 𝑆𝐼𝑍𝐸𝑗𝑡, which is defined as the market value of equity at the end of the fiscal-year, is both positive and significant at the 1% level. This indicates that larger firms from our sample give greater incentive compensation in comparison to guaranteed income to their CEOs. This is in accordance with our theoretical expectation through research done by Frydman & Jenter (2010).
Table 1
Determinants of incentive – Ratio
1992 - 2004
𝑅𝐴𝑇𝐼𝑂𝑗𝑡=∝0+∝1× 𝑆𝐼𝑍𝐸𝑗𝑡+∝2× 𝑅_𝐼𝑁𝑉𝑗𝑡+∝3× 𝐴𝑂𝐶𝑗𝑡+∝4× 𝐶𝐹𝑂𝑗𝑡+∝5× 𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡+∝6 × 𝑇𝐼𝑀𝐸 +∝7× 𝑆𝑂𝑋 +∝8× 𝑅_𝐼𝑁𝑉𝑗𝑡× 𝑆𝑂𝑋
RATIO Coef. t-stat
SIZE 0.0001257 17.86*** R_INV 25.24848 4.60*** AOC 0.0013648 12.35*** CFO -0.0003851 -9.54*** GROWTH 2.664852 6.07*** TIME 0.2465222 3.21*** SOX -1.935089 -3.18*** R_INV*SOX -21.35516 -1.32 Intercept 2.485174 5.09*** N 4747 R-SQUARE 0.1252 F-Value 84.75
25 The coefficient for 𝑅_𝐼𝑁𝑉𝑗𝑡, which is defined as the proxy for risky investments of executives and measured by the capital expenditures as a proportion of sales, is both positive and
significant at the 1% level. This indicates that firms with greater incentive compensation in comparison to guaranteed income, invest relatively more in capital expenditures. This result agrees with our theoretical construct. Ryan & Wiggins (2002) show that when managers are compensated for long-term cash flows, there will be more investment in R&D and capital expenditures(the long-term horizon).
The coefficient 𝐴𝑂𝐶𝑗𝑡 is defined as all other compensation besides bonus, salary and options granted and is both positive and significant at the 1% level. This indicates that firms with greater incentive compensation in comparison to guaranteed income, receive relatively a higher amount of other forms of compensation. Companies which use incentive compensation also use a higher amount of other forms of compensation for motivation purposes and the alignment between shareholders and agents, because other forms of compensation are another form of incentives (Carter, Lynch & Zechman, 2009).
The coefficient 𝐶𝐹𝑂𝑗𝑡which is defined as cash flow from operations is negative and significant at the 1% level. This is a surprising result, because this indicates that CEOs gain greater incentive compensation in comparison to guaranteed income in periods of lower performance. This seems illogical, but it could be the case that there is some delayed decision making which causes the incentive compensation to grow when cash flow from operations goes down.
The coefficient 𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡 which is defined as growth in net sales, is both positive and significant at the 1% level. This suggests that CEOs of growing firms are provided with more incentive compensation in comparison to guaranteed salary, which is supported by previous literature (Jensen, 1986).
The coefficient for 𝑇𝐼𝑀𝐸 is both positive and significant at the 1% level. Prior research shows an increase in incentive/option compensation over time, which supports this result (Murphy, 1999).
The coefficient of the dummy variable 𝑆𝑂𝑋 is negative and significant at the 1% level. This result indicates that firms responded with a lower ratio of incentive to guaranteed
compensation for executives after the passage of SOX. This supports our hypothesis that environmental changes on the composition of compensation induces higher guaranteed
income (Basu, Lal, Srinivasan & Staelin, 1985). It could however be the case that this effect is not solely to the passage of SOX. The passage of SOX was a result of many corporate
26 scandals which gave a lot of bad publicity for high earning CEOs, which could have had an effect on the ratio of incentive to guaranteed compensation (Ball, 2009).
The coefficient of the interaction variable 𝑅_𝐼𝑁𝑉𝑗𝑡× 𝑆𝑂𝑋 is positive, but not
significant at all. This may conclude that there is no relation between the passage of SOX and the amount of risky investments made. We elaborate on this issue when discussing table 2.
4.2 Executives’ risk-taking investment decisions
Table 2 present the results of the adjusted estimating equation (2). Tenure is not included in this adjusted equation to avoid losing 2/3th of our observations. Capital expenditures (CAPX) is used as the proxy for risky investments. Most variables in the regression are statistically significantly associated with the decision making process of risky investments, except for the variables 𝑇𝐼𝑀𝐸, 𝑆𝑂𝑋 and 𝑅𝐴𝑇𝐼𝑂𝑗𝑡× 𝑆𝑂𝑋.
The coefficient for 𝑆𝐼𝑍𝐸𝑗𝑡, which is defined as the market value of equity at the end of the fiscal year, is both positive and significant at the 1% level. This indicates that larger firms from our sample invest relatively more on capital expenditures and therefore focus more on riskier/longer-term horizon investments than smaller firms. This result agrees with the
Table 2
Determinants of incentive – R_INV
1992 - 2004
𝑅_𝐼𝑁𝑉𝑗𝑡 =∝0+∝1× 𝑆𝐼𝑍𝐸𝑗𝑡+∝2× 𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡+∝3× 𝐶𝐹𝑂𝑗𝑡+∝4× 𝐴𝐺𝐸𝑗𝑡+∝5× 𝑅𝐴𝑇𝐼𝑂𝑗𝑡+∝6 × 𝑇𝐼𝑀𝐸 +∝7× 𝑆𝑂𝑋 +∝8× 𝑅𝐴𝑇𝐼𝑂𝑗𝑡× 𝑆𝑂𝑋
R_INV Coef. t-stat
SIZE 1.26e-07 6.16*** GROWTH 0.0434473 41.69*** CFO 2.29e-07 2.00** AGE 0.000127 2.19** RATIO 0.0001837 4.25*** TIME -0.0001317 -0.61 SOX -0.0022238 -1.35 RATIO*SOX -0.0000678 -0.65 Intercept 0.0058694 1.83* N 4747 R-SQUARE 0.2977 F-Value 251.08
27 Schumpeterian hypothesis by Fisher & Temin (1973) which explains that size is significantly positively related to risky investments.
The coefficient 𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡 which is defined as growth in net sales, is both positive and significant at the 1% level. This suggests that CEOs of growing firms spend more on capital expenditures. The coefficient 𝐶𝐹𝑂𝑗𝑡, which is defined as cash flow from operations, is both positive and significant at the 5% level. This indicates that during times of greater performance, there are relatively more investments in capital expenditures. We would expect these results, because one can expect more investments when a firm also has the means to do so.
The coefficient 𝐴𝐺𝐸𝑗𝑡, which is defined as the age of the CEO, is positive and significant at the 5% level. This result disagrees with prior results in literature. Dechow & Sloan (1991) show that CEOs in their final years reaching retirement, impose reduced risk-taking investments decisions. This is presumably the case to increase reported earnings in their final years. Our different result could be caused by the use of capital expenditures and not R&D as a proxy for risky investments. Cohen, Dey & Lys (2004) did research into the same topic for a different industry and measure a significant negative coefficient for age and a positive non-significant coefficient for capital expenditures.
The coefficient 𝑅𝐴𝑇𝐼𝑂𝑗𝑡 is positive and significant at the 1% level. This indicates that CEOs receive a higher ratio of incentive to guaranteed compensation when investing more in capital expenditures. This result is supported by table 1, where we find the same, and by the prior research done by Ryan & Wiggins (2002).
The coefficient of the dummy variable 𝑆𝑂𝑋 is negative but not statistically significant. This result does not meet our expectations and indicates that the risky investments of firms did not diminish after the passage of SOX. It however does agree with our findings in table 1, where there was also no significant relation between risky investments and the passage of SOX. Because utility firms are a low-technology sector, there may be differences with how they react to environmental changes in comparison with industrial firms. McConnell and Muscarella (1985) did research on the market reaction due to capital expenditures by utility and industrial firms. They found that announcements of increases or decreases in capital expenditures had a significant effect on stock returns for the industrial sector. In the utility sector however, announcements of capital expenditure did not have any significant effect on stock returns. Because utility firms are less invested in research and long-term horizon
28 expenditures, it could be the case that environmental changes do no not have any effect on capital expenditures.
The coefficient 𝑇𝐼𝑀𝐸 is not significant, which suggests that capital expenditures did not change over the years. The coefficient of the interaction variable 𝑅𝐴𝑇𝐼𝑂𝑗𝑡× 𝑆𝑂𝑋 is negative, but not significant. This indicates that contradicts our expectations and our results in table 1.
Next we compare table 2 with table 3, which represents the results of our instrumental variables (2SLS) regression. Table 3 is controlled for economic changes and alternative decisions of executives due to a different compensation structure. Both tables have
approximately the same results, R-square and F-value. Only the variables 𝑅𝐴𝑇𝐼𝑂𝑗𝑡 and 𝐶𝐹𝑂𝑗𝑡 are not significant anymore.
Just like in table 2, the coefficient of the dummy variable 𝑆𝑂𝑋 is negative, insignificant and does not support our expectations that the passage of SOX lowered the amount of risky investments in utility firms. Table 1, 2 and 3 all indicate that there is no significant relation between risky investments and the passage of SOX.
We think that this is the case because utility firms are a low-technology sector. McConnell & Muscarella (1985) and Frenkel & Shefer (2005) both find significant relations
Table 3
2SLS regression – R_inv
1992 - 2004
𝑅_𝐼𝑁𝑉𝑗𝑡 =∝0+∝1× 𝑆𝐼𝑍𝐸𝑗𝑡+∝2× 𝐺𝑅𝑂𝑊𝑇𝐻𝑗𝑡+∝3× 𝐶𝐹𝑂𝑗𝑡+∝4× 𝐴𝐺𝐸𝑗𝑡+∝5× 𝑅𝐴𝑇𝐼𝑂𝑗𝑡+∝6 × 𝑇𝐼𝑀𝐸 +∝7× 𝑆𝑂𝑋 +∝8× 𝑅𝐴𝑇𝐼𝑂𝑗𝑡× 𝑆𝑂𝑋
R_INV Coef. t-stat
RATIO 0.0000164 0.04 SIZE 1.41e-07 3.26*** GROWTH 0.0440907 23.43*** CFO 1.84e-07 1.16 AGE 0.0001347 2.21** TIME -0.0000784 -0.31 SOX -0.0034119 -1.02 RATIO*SOX 0.0000853 0.22 Intercept 0.0059647 1.86* N 4747 R-SQUARE 0.2955 F-Value 248.05
29 between risky investments and operating income/stock prices for industrial firms. There is however not such a significant relation with utility firms. Our research shows the same. There is no significant relation between risky investments and the passage of SOX for utility firms, but there is a relation with industrial firms, according to Cohen, Dey & Lys (2004). There may be differences in how utility firms react to environmental changes in comparison with industrial firms. We expect that the investment decisions by utility firms are less affected by compensation and economic/environmental changes.
30
5. Conclusion
This paper tries to answer the question, what the effect of the Sarbanes-Oxley Act is on changes in the composition of the compensation of CEOs of utility firms; and how this affect their investment decisions. We based this study on one by Cohen, Dey & Lys (2004), who find a negative relation between the passage of SOX and the ratio of incentive compensation (value of options + bonus compensation) to fixed compensation (guaranteed salary) in industrial firms. They also find a decrease in risk-taking investments as a proportion of sales in industrial firms after the passage of SOX.
Cohen, Dey & Lys (2004) however totally omit utility firms in their sample.
According to Murphy (1999) there are significant differences in the total and composition of compensation between industrial and utility firms which may lead to other results than those from Cohen, Dey & Lys (2004). Other studies like those by McConnell & Muscarella (1985) and Frenkel & Shefer (2005) find significant relations between risky investments and
operating income/stock prices for industrial firms, while there is no significant relation with utility firms, which also supports differences in results.
Theory however still suggests that higher risks for managers will lead to less options granted and higher guaranteed salary, which is why we still hypothesize a negative relation between the passage of SOX and the ratio of incentive compensation to fixed compensation in utility firms (Basu, Lal, Srinivasan & Staelin, 1985). Managers who bear greater residual risk, because of regulatory changes like SOX, want to reduce the residual uncertainty of their investments by making less risky investments, which is why we hypothesize a decrease in risk-taking investments in utility firms after the passage of SOX (Ryan & Wiggins, 2002). Our sample consist of utility firms (SIC between 6000 and 6999) in the years 1992-2004. We find a negative relation at a significant level of 1% (t = -3.18) between the passage of SOX and the ratio of incentive to guaranteed compensation. This result indicates that firms responded with a lower ratio of incentive to guaranteed compensation for executives after the passage of SOX and supports our hypothesis that environmental influences on the
composition of compensation induces higher guaranteed income (Basu, Lal, Srinivasan & Staelin, 1985). It could however be the case that this effect is not solely due to the passage of SOX. The passage of SOX was a result of many corporate scandals which gave a lot of bad publicity to high earning CEOs, which could have had an effect on the ratio of incentive to guaranteed compensation (Ball, 2009).
31 There is however an insignificant relation (table 1: t = -1.32, table 2: t = -1.35, table 3: t = -1.02) between the passage of SOX and the amount of risky investments, according to all 3 regressions we conducted. These results do not match our expectations and indicate that the risky investments of firms did not drop after the passage of SOX. We suspect this to be the case, because utility firms are a low-technology sector. There may be differences with how they react to environmental changes in comparison with industrial firms, as mentioned by McConnell & Muscarella (1985) and Frenkel & Shefer (2005). Because utility firms are less based on research and long-term horizon expenditures, it could be the case that environmental changes have less effect on capital expenditures. We however did not have access to all the variables we wanted to use in our regression, which may cause some limitations.
The passage of SOX created legal procedures that were meant to reduce the incidence of negligence and fraud in the future. The major legal procedures caused for big changes, which gave much questions in the academic community. There are already several papers that document costs and benefits of SOX and we provide additional evidence of the costs of this act. There is however still much research to be done to conduct a more meaningful analysis, by looking at the influence of the media during the passage of SOX on the composition of compensation, the influences of environmental changes on risky investment behavior in low-technology firms and how high the costs of these changes in compensation and investment behavior were.
32
6. References
Ailon, G. (2011) Mapping the cultural grammar of reflexivity: the case of the Enron scandal. Economy and Society, Vol. 40, No. 1, pp. 141-166.
Ball, R. (2009) Market and Political/Regulatory Perspectives on the Recent Accounting Scandals. Journal of Accounting Research, Vol. 47, No. 2, pp. 277-323.
Basu, A.K., Lal, R., Srinivasan, V. & Staelin, R. (1985) Salesforce Compensation Plans: An Agency Theoretic Perspective. Marketing Science, Vol. 4, No. 4, pp. 267-291.
Bohi, D.R. & Burtraw, D. (1992) Utility Investment Behavior and the Emission Trading Market. Resources and Energy, Vol. 14, No.1, pp. 129-153.
Campbell, T.L., Umanath, N.S. & Ray, M.R. (1993) The Impact of Perceived Environmental Uncertainty and Perceived Agent Effectiveness on the Composition of Compensation Contracts. Management Science, Vol. 39, No. 1, pp. 32-45.
Carter, M.E., Lynch, L.J. & Zechman, S.L.C. (2009) Changes in bonus contracts in the post-Sarbanes–Oxley era. Review of Accounting Studies, Vol. 14, No. 4, pp 480-506.
Charoenwong, C. & Chung, K.H. (1998) Investment opportunities and market reaction to capital expenditure decisions. Journal of Banking & Finance, Vol. 22, No. 1, pp. 41-60.
Claypool, G., Tackett, J. & Wolf, F. (2004) Sarbanes-Oxley and Audit Failure A Critical Examination. Managerial Auditing Journal, Vol. 19, No. 3, pp. 340-350.
Cohen, D.A., Dey, A. & Lys, T.Z. (2004) The Sarbanes Oxley Act of 2002: Implications for Compensation Structure and Risk-Taking Incentives of CEOs. Working
