Pay-performance sensitivity : evidence from before and after the financial crisis

Pay-Performance Sensitivity:

Evidence from before and after the Financial crisis

Winston Limon 0515876

MSc Accountancy & Control, variant Accountancy Amsterdam Business School

Supervisor: Dr. Bo Qin

Co-assessor: Dr. Alexandros Sikalidis 2013-2014

Abstract.

In this thesis I examine the pay-performance relationship of the companies in the S&P 500 before and after the financial crisis of 2008. I also highlight other factors that influence the pay-performance relationship such as firm size, corporate governance and managerial power. I find mixed results, but enough empirical evidence to state that not all components of incentive-based contracts align CEO and shareholders interest effectively and efficiently. However, the results do indicate that the hybrid performance variable Tobin’s Q, has a stronger positive relation with total executive compensation and its components after the financial crisis. I also find an increase in average equity-based compensation, suggesting that firms are stepping away from cash-bonuses. Perhaps they believe that equity-based

components aligns CEO and shareholder interest more effectively. Overall this study gives evidence that support both the efficient contracting theory and the managerial power theory. This supports the claim that while the two theories are contradictive, they are not exclusive.

Table of Content 1. Introduction 1.1. Background Information 1.2. Motivation 2. Corporate Governance 2.1. Agency Problems 2.2. Governance Mechanisms 2.3. Board Structure 3. Executive Compensation 3.1. Executive Compensation design

4. Literature review &Hypothesis Development 4.1. Literature review 4.2. Hypothesis development 5. Methodology 5.1. Data 5.2. Variables 5.3. Research method 6. Regression Results 6.1. Descriptive statistics 6.2. Correlation matrix 6.3. regression results 7. Conclusion 8. Bibliography 1. Introduction

The pay-performance relationship is an issue of ongoing interest. Top executives in public firms earn relatively high compensation to align their own interest with those of the

shareholders. Next to a base salary most executives also earn incentive compensation, such as annual bonus and equity-based compensation based on their performance. While this

incentive compensation creates incentives for the CEO to increase effort, it also creates incentives for the CEO to manipulate the financial reporting to reach target set for bonus. This leaves investors wondering if shareholder value s truly maximized paying CEOs certain compensation components.

Prior academic research, examining pay-performance relationships comes up with diverse results. Aggarwal & Samwick (2003) find a positive pay performance relationship during their study of public firms in the U.S. . Hall & Liebmann (1998) also finds a positive relation

between pay and performance. However, Duffhues & Kabir, (2008), who study the pay-performance relation of public companies in the Netherlands between 1998 and 2001, find a negative pay performance relationship. These mixed results can be attributed to other factors such as firms size (Allen, 1981) and corporate governance. Also, different researcher

examine different components of compensation, some may examine only annual bonuses while others also consider long term compensation such as stock options.

Research regarding executive compensation is typically dominated by the efficient

contracting theory and the managerial power theory. The Efficient contracting theory states that the observed level and composition of executive compensation reflects the competitive equilibrium in market for managerial talent, and that incentives are structured to optimize firm value. The Managerial power theory claims that both the level and composition of compensation are not determined by competitive market forces but rather by powerful CEO’s often working through or influencing captive board members. History shows that these two theories are not exclusive.

Lawmakers and institutional investors worldwide collaborate to make laws with the intent to reduce accounting manipulation and strengthen corporate governance. Legislation such as the SOX act has increased the fines and sanctions for CEO’s that engage in manipulation, in the U.S significantly. In the Europe, however, the sanctions are relatively lower than in the U.S. Take Holland for example, the Dutch version of the SOX act, Code-Tabaksblat, is not mandatory for Dutch public companies. Also Dutch judgment often hands out less severe punishment compared to U.S judgment in cases of misreporting.

In the case of Ahold both the CEO and the CFO of the company got a suspension and a fine, after confessing that reports from 1999-2002 were not reliable. Peter de Vries, head of the Dutch shareholders organization VEB called the fine a 'pittance” stating that the CFO had a personal wealth of €43 million and was only given a penalty of €225.000. Prosecution wanted a jail sentence of 14 months, but both executives left with only a suspension. De Vries was quoted saying "this judgment sends a signal to managers that no matter what they do, the risk of a heavy punishment is minimal. In the United States, a conviction on the same facts would have led to a prison term of more than 10 years. This is Holland at its smallest". De Vries does have a point, as their colleague executives in the U.S received higher penalties and prison sentences up to 7 years from the U.S grand jury. These relatively lower sanction and

fines in the Netherlands could create more incentives for (powerful) CEO's to manipulate financial reporting, affecting the pay-performance relationship.

Next to corporate governance, firm-size also influences the pay-performance relationship as CEO's of larger companies earn a higher compensation than CEO's in smaller firms (Toshi et al. 2000). Because investors are critical about the high compensation levels being paid, I examine the S&P 500, consisting of the 500 biggest firms in the U.S. If Toshi’s findings are accurate, the executives of the S&P 500 will have the highest compensation in the U.S. While relatively little research has been done on the effect that corporate governance has on the pay-performance relationship, even less research has been done on the effect that the financial crisis had on the pay-performance relationship in the U.S. The financial crisis started in 2008 and most public trading companies saw a decline in company performance, leading to brutal financial results for investors. While the pay-performance relationship in the biggest public firms has always been an issue for debate, investors and lawmakers started placing more emphasis on creating a stronger pay-performance relationship once the financial crisis started in 2008.

What effect did the financial crisis have on the pay-performance relationship of the S&P 500 companies??

In the next paragraph I will present the motivation. In section 2 and 3 corporate governance and executive compensation are described , followed by the literature review and hypothesis development in section 4 and the methodology in section 5. The results will be presented in section 6, followed by the conclusion.

1.2 Motivation Academic Relevance

The Pay-performance relationship has been a topic that researchers have found different results for. The diversity in results can be explained by several factors; however we believe that CEO power and firm size are very important factors when examining pay-performance

relationship. Relatively little research has been done examining the factors that influence pay-performance relationship, even less research has been done regarding CEO compensation in the Netherlands. First examining the pay-performance relationship from 2004 until 2007, and then the pay-performance relation from 2008 until 2011, gives more insight off the affect the financial crisis had on the pay-performance relationship in big public firms, adding to the already existing literature.

Societal relevance

CEO’s in large companies get compensated well, their performance however doesn’t always hold up to their compensation though. Duff hues & Kabir, (2008) even found a negative relation between pay and performance, suggesting that CEO compensation is still high even if their firm performance is bad. This could be due to either weak corporate governance or very powerful CEO’s. By examining CEO power and CEO compensation in relation to firm performances, investor might be able to steer away from firms with a powerful CEO, recognizing this as a sign of weak corporate governance, and make smarter decision when choosing between companies.

2. Corporate governance

Corporate governance is a set of mechanisms and processes that help ensure that companies are directed and managed to create value for their owners, while fulfilling responsibilities to other stakeholders (Merchant, 2007). Corporate governance deals with the ways in which suppliers of finance to corporations assure themselves of getting a fair return on their

investment (Sheleifer & Vishny, 1997). In recent decades investors, regulators and lawmakers have placed more emphasis on corporate governance, due to the existence of Agency

2.1. Agency Problems

The agency theory discusses the relationship between the principle and the agent. Jensen and Meckling (1976) define the agency theory as a contract under which one person (principle) engages another person (agent) to perform some service on their behalf, which involves delegating some decision making authority to the agent. From within the agency theory streams the positive theory of agency. The positive theory views the firm as a set of

contracting relationships between shareholders and executives of public firms However, due to moral hazard, a goal divergence exists between the principle and the agent. The agency theory argues that an agent will make decision that maximize his own wealth, not necessarily that of the principle. When an agent does not act in the best interest of the principle, it’s called an agency problem.

In the case of public firms there is a separation between shareholders (principles) and executives (agents). Berle and Means (1932) state that a large public firm can have so many shareholders that no single shareholder own a significant fraction of the outstanding stock, meaning no single shareholder has enough power to control managers actions. Berle and Means also highlight an agency problem regarding on-the-job consumption. Shareholders want executives to maximize their shareholder value, but executives often enrich themselves at shareholder expense. Perks such as expensive offices, luxurious cars, 1st class plane tickets and hotel suits are just part of the on-the-job consumptions that reduce shareholder value. However, the biggest agency problem in the corporate world today is the manipulations of financial statements, in order for the agent to receive incentive compensation. An executive might intentionally misreport to make it seem that he has reached a pre-set target. Likewise, when an executive knows that the floor of the next bonus target is out of reach, he might manipulate the financial reporting to make company performance look worse than it actually is. This way the executive won’t receive a higher target the next period. Financial scandals such as Enron and Ahold are proof of the existence of this agency problem within the corporate world.

2.2. Governance mechanisms

Governance mechanisms are implemented in firms due to agency problems, they are used so the strategy and decision making better reflect the interests of the shareholders. Governance mechanisms can be distinguished between internal- and external governance mechanisms.

External governance mechanisms

- Corporate takeovers

The investment world consists of individuals and companies that either buy ownership in or take over an undervalued company. After a corporate takeover,

management of the undervalued company is often replaced. It is usually assumed they are responsible for the undervalued firm underperforming. Knowing this, executives will increase effort to make sure the firm does not underperform, making a takeover more unlikely.

- Regulation & Laws

Lawmakers and regulators worldwide often collaborate to make new laws with the intent to reduce accounting manipulation and strengthen corporate governance. Legislation such as the SOX act was created to increase the cost of misreporting executives of public firms in the U.S. The cost of misreporting for executives could consist of suspensions, fines, bonus crawl backs, prosecution and reputation loss.

Internal governance mechanisms

- Board of directors

Through the board of directors shareholders can monitor the decision making of executives. This reduces the asymmetric information, giving shareholders more belief the firms are being operated in ways that will maximize shareholder wealth. More about board structure in the next paragraph.

- Ownership structure

While I earlier stated that a public firm can have many shareholders, the last decade a mayor shift has taken place towards ownership concentration. Large blocks of the outstanding shares of many modern firms are in hands of institutional investors, who discipline ineffective executives of members of the board of directors. In this regard

concentrated ownership functions as a better governance mechanism, because diffused ownership leads to weaker monitoring of executive decision making. - Executive compensation

Firms try to use compensation as a governance mechanism that seeks to align the interest of shareholders and executives. This compensation can exist of salaries, bonuses, stocks, options and pension plans. Shareholders assume that the

compensation will align the interest of the executive with that of the shareholders, leading to an increase in shareholder value. However, shareholders must stay cautious of the fact that long term compensation plans could be subject of managerial

manipulation. More on executive compensation in the next chapter.

2.3 Board structure

In today’s investment world the structure of the board of directors can give signals to

shareholders regarding corporate governance within the firm. Typically there are two types of board structure, the two-tier regime and the one-tier regime. In the Netherlands a statuary two-tier regime is used since 1971, which is a variant of the continental European model. Anglo-Saxon countries, such as the US and the UK, often use a one-tier model.

Fama and Jensen(1983) state that the effectiveness of the board models is associated with the formal independence of the boards, and identify two activities of the board:

- Decision management, referring to the tasks of executive directors to implement strategic decisions.

- Decision control, referring to the tasks of non-executive directors to ratify and monitor executive decisions.

One-tier model

Within a one-tier model executive and non-executive directors operate in one board. The mixture of executive- and non-executive directors tends to differ among firms. The role of the CEO within the board also differs. Some one-tier have a CEO that is as chairman of the board, others have a board leadership structure that separates the CEO and chairman of the board. The one-tier board model has two major advantages:

1. A more close cooperation between the non-executive directors and the executive directors

2. A more direct and closer supervision of the executive directors.

These two advantages should lead to a more effective and efficient managing of the company. However, if the CEO of the firm is also the chairman of the board, this might decrease the independence of the board and create too much power for the CEO.

Two-Tier Model

The two-tier model was designed to separate the executive function from the monitoring function, consisting of a supervisory board and a management board. The supervisory board consists of only non-executive directors, who could for example, represent a institutional investor or the government. The management board consists of only executive directors. Thus, the CEO has no seat on the supervisory board, making its board leadership structure completely independent of executive function of the board.

In the Netherlands the Two-Tier reform Act was introduced in 2004 to give shareholders more rights, and strengthen corporate governance by reducing the power of the supervisory board. The following rights have been granted to the shareholders:

- The right to determine the main characters of the remuneration policy - The right to adopt the annual account

- The right I dismiss the entire supervisory board

- The right to reject any nomination before the appointment of the supervisory directors.

- The right to appoint the members of the supervisory board.

3. Executive compensation

Executive compensation is used to align the interest of the executives and the shareholders of a public company, reducing the agency problem regarding executive decision making. However, in some academic literature, executive compensation plans are being described as

the biggest cause of the agency problem. Bebchuk et al. (2002) explains the “Managerial power theory”, stating that the relative balance of power between executives and shareholders is argued to influence the decision making within the firm and therefore influence the level and structure of the executive compensation. The managerial power theory states that because of the principle-agent relation, agents are in the natural position to use their discretion to set their own compensation. For example, a CEO of a firm, who is also the chairman of a one-tier board of directors, is very powerful and can effectively influence his compensation structure. From the managerial power perspective, pay structure is not a solution to the agency problem. It is seen as part of the problem, or even an agency problem in itself. While a CEO managerial power is often decide by corporate governance factors such as percentage of inside and outside board members and independence of the chairman of the board (Larcker & Tayan, 2011), firm size also influences the power a CEO may have in a company. The larger the firm, the more asymmetric information the shareholders have to deal with, the more chance a CEO has to manipulate firm’s performance. Allen (1981) calls the large corporation one of the most promising substantive research arenas for

organizational stratification, stating that corporate managers often overcompensate

themselves for their activities due to unchallenged power within larger firms. Also, Tosi et al (2000) and Murphy (1999) find evidence that executives or larger firms receive higher compensation.

Another way of viewing executive compensation is through the “Optimal contracting theory”. The theory states that compensation schemes are assumed to be designed by boards to

provide managers with sufficient incentives to maximize shareholder value. Optimal contracting theory assumes that executives suffer from and agency problem and don’t necessarily make decisions that maximize shareholders value. Jensen and Meckling (1973) argue that an optimal contract is based on efficiency arguments and is the most efficient trade-off between different types of agency costs that minimize residual losses for

shareholders. However, when considering the financial scandals in recent decades, it hard to agree with this theory.

3.1 Executive compensation design.

When examining executive compensation one must distinguish incentive compensation components from the components that are less or not related to the performance of executives. Non-incentive compensation components could be: Base salary, health care, pension rights and other contributions. However, when viewing executive compensation from the managerial power perspective, it is the incentive compensation that creates the most incentives for executive to manipulate company performance, or make decisions that do not maximize shareholder value. The following compensation components are considers

incentive compensation;

Bonus plans

Bonus plans are a common form of executive compensation. Cash bonuses are often

expressed as a percentage of the executive’s base salary. Including a bonus component in an executive compensation plan requires subjective judgment on the executive’s performance. It’s important that performance is judged based on both accounting performance measures and non-financial measures.

Murphy and Jensen (2011) illustrate the situation of an CEO with a bonus plan based on return on capital (ROC), faced with non-mutual investment projects with return of capital ranging between 5%-30%. If the cost of capital is 10%, the value maximizing decision is to take all projects earning over 10%. But if the CEO’s bonus plan is only based on ROC, he will forgo all other plans and rationally accept only the 30% ROC project, even if the project is very small. Murphy and Jensen state that by doing this he reduces the level of investment to that single project with the highest return and so maximize the return on capital.

Thus, the weight placed on non-financial measures in CEO’s bonus plans is crucial for maximizing shareholder value. Itter et al. (1997) state that non-financial measures provide information about managerial actions not captured by financial measures. However, they also state that non-financial measures can easily be manipulated and are rarely subject to public verification; enabling CEO’s to inflate their compensation.

While annual bonuses are intended to incentive executive action towards used performance measures, bonus plans may also elicit non-congruent behavior such as accounting

manipulation and real economic manipulation:

- Postpone performance to next year if floor is out of reach - Accelerate performance if close to incentive zone

- Postpone performance to next year if you exceed cap.

Murphy and Jensen (2011) quoted a CEO who participated in a bonus plan based on return on equity with an upper threshold of 15%, saying “I would have to be the stupidest CEO in the world to report an ROE of 18%. First, I wouldn’t get any bonus for any results above the cap. Second, I could have saved some of our earnings for next year. And third, the board of

directors would increase my target performance for next year”. The CEO’s comments are evidence of the existing agency problem and create the belief that the Managerial power theory is superior to the Optimal contracting theory.

Stock ownership

As stated earlier, executives are often more interested in on-the-job compensation and

expensive perks, than maximizing shareholder value. One way to reduce this agency problem to some degree is by including a Stock component to the executive compensation plan. This way executive compensation is more closely related to the stock returns of the company. Jensen and Meckling (1976) argue that a manager who owns 100% of the shares of a company will consume on-the-job perks (luxurious offices, pt projects) to a point that

marginal utility from one-dollar expenditure on consuming perks is offset by marginal utility from one-dollar decrease in wealth. However, if the same manager was to sell 5% of equity claims, the manager would consume more perks. He would enjoy a one-dollar marginal utility, but only experience a 0.95 dollar decrease in wealth for each dollar spent on perquisites. As the owner-manager fraction of equity falls, the manager's fraction on the residual claims falls, creating more incentives for the manager to consume perquisites. Therefore a manager with zero percent ownership will consume more perquisites, given that

little to none of his wealth is tied to the performance measure stock price. Once his fraction of equity increases, the manager will be less inclined to consume perquisites.

This is not only the case with perquisites consumption, but with all actions and decisions made by the manager that affect shareholder value. If the separation of ownership and control creates a moral hazard, the separation can be reduced by increasing executive ownership of the company.

Stock Options

Stock options provide executives with right to buy shares at a pre-specified exercise price. The shares can be bought after the vesting period of typically 3-5 years. Most options are granted at the money, meaning that the exercise price equals stock price at grant date. Vesting of option could also be contingent on achievement of performance target.

Stock options are the most important component of monetary incentives for U.S. executives. The strongest incentives follow from executive’s equity portfolios, relative to equity grants. Prior research argues that rewarding executive stock options creates a stronger relation between executive compensation and stock returns. With the proper contract structure, options give a stronger stimulus than the stimulus of the stocks themselves. Executives will increase effort to improve company performance, because they don’t want stock prices to decrease.

However, including stock options in an executive compensation plan is not a guaranteed solution for the agency problem. Stronger yet, researcher argue that stock options create the most incentive for executives not only to manipulate earnings, but also to adopt in more risky investment and debt policies. Coles et al. (2006) examines executive decision making

regarding risky investments and debt policies, and highlight 2 risky policies executives take: 1. Reallocating investment dollars from tangible assets to intangible assets. This

increases risk as investments in PPE are considered less risky compared to investments in R&D.

2. Less diversification, increase focus and increased leverage.

Due to such policies many believe the claim that excessive risk taking caused the financial crisis and that the structure of executive compensation contracts regarding options

and market expectations. Larker & Tayan (2011), however, argue that excessive risk is not a well-defined term. CEO’s are responsible for corporate strategy and this forces them to pursue investments that are risky, in the sense that their future payout is unknown in advance. They state that there is no bright rule that distinguishes excessive risk from acceptable risk.

Long Term Incentive Plans

These plans are often based on stock and options. A performance target is set a few years in advance. If the executive reaches this pre-set performance target, he will be granted the stocks or options. Most options have a vesting date of 3-5 years, creating long term incentives for the executive.

If stock grants are based on performance, restricted stock represents a combination of compensation for prior performance, and incentives to increase future stock price. This creates less of a divergence of interest between executives and outside investors since executives also bear the cost and benefits of the decision making to a greater degree.

4. Literature Review & Hypothesis development 4.1. Literature review

Not much academic research has been done regarding the pay-performance relation;

however, the results have been mixed and are not pointing in one direction. Two widely cited empirical studies are those of Murphy (1999) and Core et al. (2003). They study the

relationship between CEO compensation and shareholder wealth in the US and find evidence that the relationship is positive and that equity-based incentives are the driving force behind it. Aggarwal & Samwick (2003) also find significantly high pay-performance sensitivity when conducting a study in the United States between 1993 and 1997; the high pay-performance sensitivity found in the study, can be contributed almost entirely to executive stock options. With regards to the U.S. there are a few studies that do not find a positive relation when examining CEO compensation and firm performance. Brick et al. (2006) find that there is a negative relationship between excess executive compensation and firm

performance. Core et al (1999) also finds a negative relationship, documenting that excess CEO compensation has a significant negative association with subsequent stock return and operating performance.

Empirical studies done outside of the U.S also show mixed results. The studies that do find a positive relationship have a very low sensitivity. Conyon and Murphy (2000) and Buck et al. (2003) both find relatively weak pay-performance sensitivity in the U.K. With regards to empirical research done in Asia, Kato and Kubo (2006) find positive pay-performance relationship while examining CEO compensation of listed and non-listed Japanese firms. Firth et al. (2006) finds a low, but positive pay-performance sensitivity for Chinese listed companies. However, other studies done outside of the U.S. show a negative

pay-performance relationship. Kabir and Duffheus (2008) find a negative pay-pay-performance relation when examining the belief that executive pay should reflect firm performance in the Netherlands from 1998-2001. They actually find a negative correlation between executive pay and corporate performance. Fernandes (2008), on the other hand, does not find any relationship between CEO compensation and firm performance when analyzing a sample of listed companies in Portugal. Other studies that could not identify a relationship between firm performance and CEO compensation include Miller (1995) and Jeppson et al. (2009).

One explanation for the mixed results could be that the researchers all use different

components of executive compensation when conducting the research. Also salaries differ at different hierarchy levels, some studies might examine all executives within a company and some only the CEO and CFO. Also salaries differ worldwide. Either way, these inconsistent results of empirical studies done both in- and outside of the U.S., raises many concerns regarding the role of compensation as a cure for agency problems. These concerns would only increase when the financial crisis of 2008 arrived. The introduction of the Dodd-Frank Act was a direct result of these concerns.

The Dodd Frank Act was signed into federal law by President Barack Obama on july 21st 2010, passed a response to the financial crisis of 2008. It sought to bring the most significant changes to financial regulation in the United States since the regulatory reform that followed the great depression. A variety of critics have attacked the law, some arguing that it was not enough to prevent another major crisis or bailout, while others argue the act went t o far n restricting several financial institutions. While the Dodd-Frank act affects almost every aspect of financial regulation in the United States, I focus only the rules created regarding executive

compensation and corporate governance. The Dodd-Frank Act instructed the SEC to enforce the following regulations regarding executive pay (Murdock.,2011):

- Say on Pay

Shareholders will be asked to approve the company’s executive compensation practices in a non-binding vote every 3 years. Furthermore , firms are required to disclose, while shareholders are asked to approve any golden parachute payments in connection with mergers, tender offers or going-private transactions.

- Claw-backs

Companies are required to implement and report policies for recouping payments to executives based on financial statements that are subsequently restated. This rule applies to both current and former executives, and any payments made in the three years period preceding the restatement.

- Compensation committee independence

Publicly traded companies are required to have a compensation committee comprised of solely outside independent directors.

- Proxy access

The Dodd-Frank Act authorizes the SEC to issue rules allowing certain shareholders to nominate their own director candidates in the company’s annual proxy statements. While empirical research regarding pay-performance relationship before the financial crisis is limited, the empirical research done after the financial crisis (regarding pay-performance sensitivity) is almost non-existing. Even less literature can be found regarding the affect the introduction of the Dodd-Frank Act had on the pay-performance relationship.

Vemala et al. (2014), however, did conduct a study while using a sample consisting of Fortune 500 firms. They found that the financial crisis had a small but significant effect on CEO compensation. They also find a significant positive relationship between firm

performance and CEO compensation both pre- and post-financial crisis. After the financial crisis cash compensation decreased significantly, while equity-based compensation increased.

Yang et al (2014) also examined the effect the financial crisis had on CEO compensation. They include S&P data from 1992 to 2011. Analyzing CEO total compensation, total cash compensation (Salary + Bonus) and CEO equity compensation. They find the

pay-performance relationship demonstrates different patterns before and after the crisis. Before the crisis each component of compensation had a significantly positive relationship with the accounting-based firm performance measures (ROA) and the stock-based firm performance measure (TSR). After the crisis they find that there is still a positive pay-performance relationship, however, there exist a negative relationship between total CEO compensation and stock-based performance. While total CEO compensation increased after the financial crisis, overall stock-based performance measure declined. These results suggest that incentive-based controls were not effective compensation tools after the crisis.

4.2. Hypothesis development

Duffhues, P., Kabir, R., (2008) argue that once CEO's receive sufficient compensation, they are assumed to increase effort and improve company performance. The pay-performance relationship is thus expected to be positive. During the financial crisis most firms worldwide endured losses, which created more political heat for CEO's regarding their compensation. Well aware of the financial situation the investment world was in; investors became more cautious and became more curious about the influence of incentive compensation on the shareholder value. With firm performance becoming so crucial, investors started placing more emphasis on creating a positive pay-performance relationship during the crisis. Due to the shift towards concentrated ownership, and the introduction of “Say-on-pay” at the annual meetings, institutional investors now have more power to influence decision making and reduce CEO power. It should be noted that the “Say-on-pay” voting is not binding, and that a firm can still pursue their pay arrangements, even at the dislike of an institutional investor. However, one could make a case that such actions could signal the wrong message to the stock market leading to a decrease of the stock price. Executives at publicly traded firms should be aware and cautious of these signals.

A powerful institutional can also apply pressure on the CFO and ask for a second opinion regarding the financial statements. If the Institutional investor suspects weak corporate governance or manipulation, such a powerful investor can request that a joint audit be done. This increases the chance that both intentional and unintentional misstatements will be found during the audit. The conservative theory states that auditors will recognize a loss quicker,

than they recognize a profit (Watts, R.L., 2003). This means that a joint audit would probably lead to a decrease of net-income, eventually leading to a decrease in bonuses the company has to pay to its executives. This will especially be the case for firms where executives have barely reached their pre-set target to receive a bonus. These firms are very suspect to

misreporting. Hiring a second auditor with a conservative demeanor, could help decrease income, leading to a situation where the firm has to pay less bonuses. This increases shareholder value.

Given the financial crisis, the switch from diffused ownership to concentrated ownership and the introduction of the Dodd-Frank Act, I expect to find a stronger positive pay-performance relationship during the years after the financial crisis, than the years before the financial crisis. This leads to the following three hypotheses:

H1: A stronger positive relationship exists between Total CEO compensation and firm performance from 2008- 2012, than from 2003- 2007 for S&P 500 companies.

H2: A stronger positive relationship exists between Cash-based CEO compensation and firm performance from 2008- 2012, than from 2003- 2007 for S&P 500 companies.

H3: A stronger positive relationship exists between Equity-based CEO compensation and firm performance from 2008- 2012, than from 2 2003- 2007 for S&P 500 companies.

5. Data and research method 5.1 Data

In this experiment I used 2 samples from the S&P 500 data set found in Compustat and ExecuComp. One set from 2003 - 2006 (sample prior to the financial crisis), and the other from 2009 - 2012 (sample after the financial crisis). The executive compensation is measured at time t, while the firm performance variables are measured at t-1. This method is used in prior studies (Firth et al., 2007), indicating that executive compensation is a direct result of previous firm performance, giving more reliable results. Thus firm performance measures will be retrieved from Compustat for the years 2003-2007, and 2009-2012. Data regarding executive compensation will be retrieved from ExecuComp for the years 2004 – 2008, and 2010 – 2013.

After retrieving the data for the years prior to the financial crisis from Compustat, I have a total of 1320 observation from 332 companies. However, I only found 1214 observation regarding executive compensation after retrieving the data from ExecuComp. Thus, 106 observations were removed from the sample. Also, companies missing data for more than one variable are filtered out, leading to more reliable results and a higher quality of data. This led to the removal of 532 firm performance observations, leaving us with a dataset of 682

observations. Table 5.1. gives a good overview of the data available. With regards to the dataset for the years after the financial crisis, I retrieved 1650 observations from ExecuComp. However, I only retrieved 1436 observation regarding firm performance from Compustat, forcing me to remove 214 observations from the dataset. After that I removed 351

observations missing more than one variable from the dataset. This leaves me with a total of 1085 observations.

Table 5.1 Sample selection procedure

S&P 500 Data Number of observations

Compustat Firm performance data 1320

Excluding observations with no CEO compensation Data 106 Excluding observations with 1 or more missing variables 532

Total observations 2003-2006 682

S&P 500 Data Number of observations

ExecuComp Executive compensation data 1650

Excluding observations with no Compustat Firm performance

data 214

Excluding observations with 1 or more missing variables 351

Total observations 2009-2012 1085

Though public firms in the U.S. are required to disclose compensation data regarding CEO’s, not all data seems to be available in ExecuComp. Likewise, not all data regarding firm performance is available in Compustat. Furthermore, all variables used in this study have been winsorized by replacing the top and bottom 1% with the most extreme, but acceptable value.

5.2 Variables

In this study I will be using the executive compensation as the dependent variable. Three hypotheses have been constructed in this study that will be measured empirically. In this section the dependent-, independent and control variables will be described.

The regression looks as following:

Payi,t = ß0 + ß1*Perfi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis+ε

Dependent variable: executive Compensation

The amount of compensation the CEO received at time t will be expressed with Payi,t, the

natural logarithm of executive compensation to adjust for standard normal distribution and to mitigate the influence of outliers. Several regressions will be run with the different

components of executive compensation. In this study the following compensation components are distinguished:

- Cash Compensation

Consisting of a base salary and short term cash bonuses, the natural logarithm of cash compensation will be expressed with the variable LN_CC.

- Equity based Compensation

Restricted stock grants and stock options measured at fair value. The natural logarithm of CEO equity based compensation will be expressed by the variable LN_EQB

- Total Compensation

Consisting of salary, bonus, restricted stock grants and the value of options. The natural logarithm of total CEO compensation will be expressed by the variable LN_TC.

Independent variables: Firm performance

The following criteria will be used when evaluating alternative performance measures: − Noise: to what extent can the measure be influenced by other external factors − Sensitivity: to what extent can executives influence the performance measure

− Congruence: to what extent does an increase in measure actually lead to an increase in firm value

− To what extent is measurement susceptible to manipulation?

These factors can either increase or decrease the prevalence- and reliability levels firm performance measures.

In an efficient market, stock price should reflect future cash-flows and risk incorporated in the discount factor. An example of a capital-based measure is TSR, which is defined as the change of share price during the year, plus the dividend payouts divided by the share price at the start of the year. Capital-based measures are forward looking, incorporating anticipated effects of current actions on future cash flow. One disadvantage is that capital-based measurements may reward executives for anticipated instead of realized performance.

Executives can also manipulate the measurement, by influencing stock price through strategic disclosure of positive or negative information to the capital market. Such manipulation makes capital-based measures less reliable. Another disadvantage is the impact of external factors rendering the measurement relatively noisy e.g. economic climate. Capital-based

measurements, however, are relatively congruent measures if current stock price is an

adequate reflection of true, economic value of the firm. And while capital-based measures are not sensitive for the actions of individual lower-level employees, however, the sensitivity increases for high ranked executives. (Merchant & Van der Stede., 2012).

Accounting-based measures have their own set of pros and cons. The biggest disadvantage, however, their weak congruence level. Due to the backwards looking nature of accounting-based measures, reflecting economic impact of decisions made in the past. Another

congruence concern stems from the conservative nature of accounting, which can lead to poor matching of current investment with future benefits. Furthermore, accounting-based

objectively. These congruence issues reduce the reliability of the accounting-based measures. However, accounting regulations and verification by auditors lower the susceptibility to manipulation compared to capital-based measures. This adds value to the use of accounting-based measures and increases the reliability of the measure. Another advantage is that the sensitivity of accounting-based measures can be improved by rewarding profit at the

appropriate level. Also, accounting-based measures are less susceptible to noise compared to capital-based measures, increasing the relevance of the measurement.(Merchant & Van der Stede., 2012).

Due to the existing advantages and disadvantages of accounting- and capital-based, firm performance will be measured using an accounting based measure, a capital-based measure and a measure that is a combination of the two. This is consistent with prior research conducted. The accounting-based measure is the Return on Assets (ROA), which is defined as the ratio of net income to average total assets. Including ROA as a performance measure is consistent with prior research done (Kabir & Duffhues., 2008 and Core et al., 1999 ). The capital-based measure is the total shareholder return (TSR), which was used by Kabir & Duffheus (2008). The last firm performance measure is the Tobin’s Q which is a hybrid of a capital- and accounting-based measure, defined as the ratio of the sum of market value of equity and the book value of debt to book value of total assets. The Tobin's Q has also been used in prior research conducted (Chung & Pruitt, 1994; La Porta, Lopez-de-Silanes & Shleifer, 1999)

Control Variables

In this study the following confound effects will be controlled for:

- Firm Size

Several studies find that firm size has a large effect on the on the amount of

compensation an executive receives. Tosi et al.(2002) find that for more than 40% of the variance in CEO pay can be attributed to firm size. The firm size is defined as the market value of the total assets of the firm. The variable LN_Size can be defined as the natural logarithm of firm size or total assets. This is done to control for outliers. Size was also been controlled for in prior research conducted by Hall & Lieberman (1998).

- Price volatility

The second control variable is volatility, defined as the variance of stock returns during a year. Companies operating in a very volatile market might pay more compensation, because executives want to be compensated for the higher risk. The price volatility will be expressed by the variable Vol. We therefore control for the effects of price volatility, consistent with prior research done by Coles et al. (2004).

- Leverage

Leverage can be defined as the ratio of total debt to asset value. Higher debt can lead to an increase firm risk, which in turn could lead to CEO compensation. Leverage will be expressed by the variable Lev. (Lippert & More.,1994; Lippert & Porter., 1997)

- Executive Age

The last control variable is the executive specific variable Age. This variable will be used because the older the executives are, the higher their compensation is expected to be. Executive age has also been controlled for in prior studies ( Coughlan & Ronald Schmidt., 1984; Mehran., 1994), and will be expressed by the variable Age.

Dummy and interactive variables

- Time

Time is the dummy variable that indicates if a certain observation belongs to the post financial crisis data, or the pre financial crisis data.

- Fin_crisis

This is the interaction between the dummy variable Time and the firm performance variable. This variable show the effect the financial crisis had on the pay-performance

sensitivity. If the coefficient is positive and significant, the pay-performance sensitivity is higher after the financial crisis.

5.3. Research Method

I used multiple regression analysis to conduct the experiment in IBM SPSS. The nine following models were used to run the regression analysis:

Total Compensation

Model 1a:

Abbreviation Definition

LN_Payi,t Natural logarithm of executive compensation LN_TC Natural logarithm of total executive compensation LN_CC Natural logarithm of cash-based compensation LN_EQC Natural logarithm of equity-based compensation

Firm Performance

Return on assets ROA Net Income / Total Assets Total share holders return TSR (P1-P0+Div)/P0

Tobin's Q Q (MVE+BVD)/(BVE+BVD)

Control Variables

Firm size LN_Size Natural logarithm of size Price volatility Vol Variance in f irms annual stock price

Leverage Lev Total debt / Total assets

Executive age Age Age of CEO

Dummy Variables

Time Time Dummy variable indicating post- or pre crisis data Financialcrisis Fin_crisis Interaction between dummy and performance variable

Executive Compensation Executive compensation Total Compensation Cash-based compensation Equity-based compensation

LN_TCi,t = ß0 + ß1*ROAi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

Model 1b:

LN_TCi,t = ß0 + ß1*TSRi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

Model 1c:

LN_TCi,t = ß0 + ß1*Qi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis+ ε

Cash-based Compensation

Model 2a:

LN_CCi,t = ß0 + ß1*ROAi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

Model 2b:

LN_CCi,t = ß0 + ß1*TSRi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

Model 2c:

LN_CCi,t = ß0 + ß1*Qi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

Equity-based Compensation

Model 3a:

LN_EQCi,t = ß0 + ß1*ROAi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

LN_EQCi,t = ß0 + ß1*TSRi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

Model 3c:

LN_EQCi,t = ß0 + ß1*Qi,t-1 + ß2*Sizei,t-1 + ß2Voli,t-1+ ß2*Levi,t-1 + ß3*Agei,t + ß4Time+ ß5Fin_crisis +ε

In model 1a-1c I exam the relationship between firm performance and total executive compensation. In model 2a-2c I examine the relationship between firm performance and cash-based executive compensation, and in model 3a-3c I examine the relationship between firm performance and equity-based compensation. All nine models will be used to estimate pay-performance sensitivity (for their respective compensation components) both before and after the financial crisis.

ß0 = Stands for the constant variable in the regression.

ß = stand for the coefficient of the dependent-, independent-, and control variables. ε = stands for the error term

Chapter 6. Results. 6.1 Descriptive statistics

The purpose of this chapter is to describe the descriptive statistics of the variables that are used on the regression analysis. First I will discuss the differences in executive compensation prior to the financial crisis and after the financial crisis. I will largely compare the statistics with that of Yang et al. (2014) and Vemala & Nguyen (2014), due to the fact that both studies examine the affect that the financial crisis had on executive compensation. They examined CEO compensation both before and after the financial crisis. Table 6.1. presents the descriptive statistics of the different components of executive compensation, firm performance measures and control variables prior to the financial crisis, while Table 6.2. presents the descriptive statistics of the dataset post financial crisis. Both tables show the Minimum, 25 percentile, 50 percentile, 75 percentile and maximum value of all variables used in the regression analysis. All variables used in this study have been winsorized by replacing the top and bottom 1% with the most extreme, but acceptable value.

Table 6.1 Descriptive statistic 2003-2009

Descriptive Statistics Percentiles

N Minimum Maximum Mean

Std. Deviation 25 50 75 TC 832 556 37762 6428.60 5532.470 3512.91 5814.99 _9605.03 LN_TotCom 832 6.32 10.54 8.4359 .83959 8.1642 8.6682 9.1700 CC 832 271 15596 1824.60 1732.681 765.64 1582.49 3162.75 LN_CCom 832 5.60 9.65 7.2311 .70884 6.6407 7.3668 8.0592 EQC 832 0 13883 1246.10 2073.738 841.15 1275.00 2200.00 LN_EQCom 431 2.41 9.54 7.2978 1.09882 6.7348 7.1507 7.6962 ROA 816 -5.58 34.59 8.6018 5.56038 4.5400 8.2800 11.5800 TSR 816 -.2485 1.5244 .233897 .2696972 .061531 .182663 .352029 Tobin's Q 816 .0164 6.424 1.372 1.2002 .471 .931 1.665 Vol 816 14.04 54.69 25.8804 8.31872 19.0400 23.3000 28.2200 Lev 816 0.00 .60 .2114 .14673 .1071 .1923 .2962 LN_Size 832 10.78 21.10 16.2867 1.39734 15.5575 16.4365 17.2522 Age 832 35 75 51.93 4.694 50.00 51.93 54.00 Valid N (listwise) 395

From 2003 till 2006 the mean of total executive compensation was equal to $6,428,600, with a standard deviation of $5,532,470. Less than 75% of all executives earn more than

$9,605,030 prior to the financial crisis with $37,762,000 being the maximum. With regards to the separate components of executive compensation, the mean for the cash components of compensation (Salary and Bonus) is equal to $1,824,600 with a standard deviation of 1,732,680, while the mean for equity based compensation is $1,246,100 with a standard deviation of 2,073,740. These values are relatively similar to the values found by Yang et al. (2014). Prior to the financial crisis Yang found mean values of $5,946,000, $1.140,750 and $1,852,500 for total-, equity-based and cash-based compensation respectively. Vemala & Nguyen (2014), however, claim that the average CEO total compensation was $ 8,628,021, while the average cash-based compensation was $1,387,179.

The mean of the natural logarithm of total executive compensation is 8.435 with a standard deviation of .84. The mean of the natural logarithm of cash-based and equity-based

compensation are respectively 7.23 and 7.29 with standard deviations of 0.71 and 1.09. The firm performance variable ROA has a mean of 8.6 % with a standard deviation of 5.6%. The independent variables TSR and Tobin’s Q show means of 19,7% and less than 1.3

respectively.

In the years following the financial crisis the average total compensation for CEO’s in the S&P 500 is $10,134,830, with a standard deviation of $6,187,280. The maximum total

compensation value is $ 56,859,170, while the minimum value is $300,930. Less than 25% of CEO’s in the S&P 500 earned more than $13 million from 2009-2012. When comparing this to the total executive compensation mean prior to the financial crisis, I find that total

executive compensation has almost doubled ($10,134,830). The standard deviation has also increased to $6,187,270 after the financial crisis. These results are consistent with the results of prior studies (Kirkpatrick., 2009 and Kavoussi., 2012), that indicate an increase in total executive compensation after the financial crisis. Vemala & Nguyen (2014), however, find an average total compensation of $8,753,325, which in their study is a relatively small increase compared to before the crisis ($8.628,021). Yang et al. (2014), on the other hand, finds an average executive total compensation of $5,551,667, which is a decrease compared to the years before the crisis.

Table 6.2 Descriptive statistic 2009-2012

Descriptive Statistics

Percentiles

N Minimum Maximum Mean

Std. Deviation 25 50 75 TC 1174 300.9270 56859.16 10134.84 6187.28 6437.07 9239.51 12890.70 LN_TC 1174 5.71 10.95 9.04 .639 8.7698 9.1312 9.4643 CC 1174 1.1270 7800. 1368.87 1052.49 947.24 1092.94 1350.00 LN_CC 1174 .12 8.96 7.04 .63029 6.85 6.9966 7.2079 EC 1174 0.0000 14868. 1974.016 2224.55 1375.78 2428.44 3782.10 LN_EQC 792 .40 9.61 7.6756 .93275 7.2268 7.7950 8.2380 ROA 1174 -12.44 28.54 7.2969 5.70 3.7350 7.1850 10.9400 TSR 1174 -.454 2.104 .19756 .312166 .01573 .15575 .32353 Tobin's Q 1174 .3360 4.484 .9921 .00078561 .0005042 .0008826 .0014503 Vol 1175 10.4100 53.64 26.16 7.844 20.3025 24.835 30.095 Lev 1174 0.0000 .7234 .240596 .15206 .118516 .220430 .325430 LN_Size 1174 14.25 21.34 16.7698 1.26027 52.00 56.00 59.00 Age 1190 37 80 55.93 5.763 15.8117 16.4937 17.4992 Valid N (listwise) 748

When further analyzing the statistics, I find that average cash compensation is $1,368, 870 with a standard deviation of $1,052,480. The average equity-based compensation is equal to $41,974,020 with a standard deviation of $ 2,224,550. The maximum value for equity based compensation is $14,868,000 which is about 10 times the mean. However, less than 25% of CEO’s earned more than $3,782,100 the years after the financial crisis. Compared to the years prior to the financial crisis, the average CEO cash-based compensation has decreased by $455.730. Vemala & Nguyen and Yang both find quite similar average cash-based compensation of $ 1,066,672,360 and $1,066,333 respectively, representing decreases compared to the years before the financial crisis.

The natural logarithm of equity-based compensation has a mean of 7.67 and a standard deviation of 0.93. The number of observations (792), however, is substantially less than the number of observations for the equity-based compensation (1174). This is due to the fact that 382 observations were registered with a value of 0, meaning that the respective CEO did not receive a bonus that year. When transforming the data to natural logarithm, values of zero are not taking into account by SPSS. The average CEO equity-based compensation is $1974,01, which is an increase of $727,91.

With respect to the firm performance variables ROA, TSR and Tobin’s Q, I find means of respectively 7,2 %, 19.7% and 0,992. Compared to the years prior to the financial crisis these values represent a decrease in firm performance, which is consistent with the results of prior studies. The lowest value of ROA in our sample is -12,44%, while the lowest value of TSR is -45,4 %.

Paired-sample t-test

For more statistical significance regarding the changes in CEO compensation, I also test both samples using a paired-sample t-test. Total executive compensation, and its components, is compared before and after the financial crisis. The results can be observed in table 6.3. It can be observed that the level of CEO cash-based compensation has decreased in the years after the financial crisis compared to the years before the financial crisis. This is consistent with the result observed in the descriptive statistics.

One possible explanation for the decrease in cash-based compensation, which consists of base-salary and bonuses, is a possible decrease in yearly bonuses being paid out to CEO’s. After the financial crisis shareholders became aware that excessive executive bonuses could create wrong incentives and placed more emphasis on reconstructing CEO compensation plans. This could also explain the increase in equity-based compensation after the financial crisis, which can also be observed in Table 6.3, leading to an increase in total CEO compensation overall. The results in Table 6.3 are significant at 1%.

Executive Compensation T Sig. (2-tailed)

Cash Compensation Before Crisis – After Crisis 6.736 .000

Equity Compensation Before Crisis – After Crisis -5.248 .000

Total Compensation Before Crisis – After Crisis -10.321 .000

Table 6.3 Paired-sample t-Test

Wilcoxon Signed-rank test

I also used the Wilcoxon signed-rank test to compare the two samples, which can be used as a substitute for the sample-paired t-test. The Wilcoxon signed-rank test can be used to determine not only whether a pair of observation differs, but also the magnitude of any difference. In table 6.4 the results can be observed.

Cash-Based Compensation Mean Rank Equity-Based Compensation

Mean Rank Total Compensation Mean Rank

Neg. Rank 456 484.82 285 374.20 271 403.59 Pos. Rank 410 376.42 471 381.10 599 449.94 Ties 4 114 0 Total 870 870 870 Z -4.532 -6.064 -10.800 Sig. (2-tailed) .000 .000 .000

The null hypothesis states that if there was no influence exerted by the financial crisis, the compensation levels of CEO’s after the financial crisis should be no different from the compensation levels prior to the financial crisis. However, the 1% significance of the results indicates that the levels of the three components of executive compensation are not similar before and after the financial crisis. The higher negative ranks and mean rank of cash-based compensation indicates that CEO receive less cash compensation after the financial crisis, compared to the years before the financial crisis. Consistent with the results of the descriptive statistics and the paired-sample t-test, CEO’s received more equity-based compensation after the financial crisis than before the financial crisis. The number of ties in the ranking of equity-based compensation is quite large, 114, due to the fact that some CEO’s do not reach their target and receive zero dollars in bonuses. Regarding the total executive compensation as positive rank of 599 and a mean rank of 449.94 indicate that the total CEO compensation has increased, even though cash-based compensation has decreased after the financial crisis. Table 6.5 presents a summary of the mean, median and P-values of the parired t-test and Wilcoxon test.

Table 6.5 Summary statistics 2003-2006 vs. 2009-2012

Variable

2003-006

(N = 444) (N = 1,297) 2009-2012 P-value

Mean Median Mean Median t-test Wilcoxon

Total compensation 6428.60 5458.62 10134.84 8857.08 0.000 0.000 Cash based compensation 1824.60 1145.50 1368.87 1086.50 0.000 0.000 Equity based compensation 1246.10 598.03 1974.02 1465.00 0.000 0.000 ROA 8.60% 7.86% 7.68% 6.54% 0.000 0.000 TSR 0.23 0.18 0.20 0.156 0.010 0.009 Tobin's Q 1.37 0.98 0.99 0.80 0.000 0.000 Volatility 25.88 24.02 26.16 24.87 0.000 0.000 Leverage 0.211 0.201 0.24 0.228 0.464 0.092 Size 21505.06 11619.61 31109.01 16259.12 0.000 0.000 Age 51.93 52 55.93 56 0.000 0.000

6.2 Correlation Matrix

The statistical relationship between the variables used in the regression analysis is presented in table 6.6 and 6.7 . When analyzing the correlation of variables, values typically range between -1 and 1, with a value of 0 meaning there is no relationship between the variables. A value of 1 represents a high positive relationship between variables, while a value of -1 represents a high negative relationship between variables. In this paragraph I will examine the correlation of the variables used in the sample prior to the financial crisis, as for the variables used in the sample after the financial crisis.

Table 6.6. presents the correlation between the variables used in the sample prior to the financial crisis. When examining the table 6.6. it can be observed that there is a significant positive correlation between the independent variables of firm performance ROA and the natural logarithms of total compensation (R= .097, P<.01) and equity based compensation ( R= .089, P< .05). However, a negative correlation can be observed between the independent firm performance variable ROA and the natural logarithm of cash compensation (R= -.083, P< .05). Furthermore the independent variable TSR shows a positive correlation with the natural logarithm of cash compensation (R= .091, P<.01). The independent variable Tobin’s Q is significantly negatively correlated with the natural logarithm of cash compensation (R= -.116, P< .01), but positively correlated with the natural logarithm of equity based

Table 6.6 Correlation Matrix. Pre Financial crisis.

Correlations

LN_TC LN_CC LN_EQC ROA TSR Tobin's

Q LN_Size Vol Lev Age

LN_TC Pearson Correlatio n 1 ,568** ,780** ,097** ,041 ,057 ,231** ,103** -,130** ,128** Sig. (2-tailed) ,000 ,000 ,004 ,228 ,094 ,000 ,002 ,000 ,001 LN_CC Pearson Correlatio n ,568** 1 ,389** -,083* ,091** -,116** ,250** ,012 ,026 ,127** Sig. (2-tailed) ,000 ,000 ,014 ,008 ,001 ,000 ,722 ,452 ,001 LN_EQC Pearson Correlatio n ,780** ,389** 1 ,089* ,071 ,105* ,166** ,159** -,119** ,016 Sig. (2-tailed) ,000 ,000 ,043 ,106 ,017 ,000 ,000 ,007 ,739 ROA Pearson Correlatio n ,097** -,083* ,089* 1 ,057 ,676** -,433** ,132** -,311** -,053 Sig. (2-tailed) ,004 ,014 ,043 ,095 ,000 ,000 ,000 ,000 ,165 TSR Pearson Correlatio n ,041 ,091** ,071 ,057 1 -,132** -,232** ,308** -,015 -,068 Sig. (2-tailed) ,228 ,008 ,106 ,095 ,000 ,000 ,000 ,656 ,075 Tobin's Q Pearson Correlatio n ,057 -,116** ,105* ,676** -,132** 1 -,480** ,290** -,432** -,147** Sig. (2-tailed) ,094 ,001 ,017 ,000 ,000 ,000 ,000 ,000 ,000

LN_Size Pearson Correlatio n ,231** ,250** ,166** -,433** -,232** -,480** 1 -,361** ,186** ,127** Sig. (2-tailed) ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,001 Vol Pearson Correlatio n ,103** ,012 ,159** ,132** ,308** ,290** -,361** 1 -,304** -,194** Sig. (2-tailed) ,002 ,722 ,000 ,000 ,000 ,000 ,000 ,000 ,000 Lev Pearson Correlatio n -,130** ,026 -,119** -,311** -,015 -,432** ,186** -,304** 1 ,051 Sig. (2-tailed) ,000 ,452 ,007 ,000 ,656 ,000 ,000 ,000 ,184 Executive' s Age Pearson Correlatio n ,128** ,127** ,016 -,053 -,068 -,147** ,127** -,194** ,051 1 Sig. (2-tailed) ,001 ,001 ,739 ,165 ,075 ,000 ,001 ,000 ,184

* means 10% significance, ** means 5% significance, *** means 1% significance.

Also the correlation between the independent variable ROA and the other independent variables used in the regression seem to be quite high. Especially the correlation between ROA and the Tobin’s Q (R= .676, P< .01) is significantly positive. This is to be expected since the total value of assets form the basis of both variables. Furthermore there is a negative significant correlation between ROA (R= -.433, P< .01) and the natural logarithm of size, which is also based on total asset value. Overall, the correlation between the independent variables does not seem to lead to multi-collinearity problems.

Table 6.7. presents the statistical relationship between the variables used in the sample after the financial crisis. It can be observed that he natural logarithm of total compensation is a positive significant relationship between the dummy variables LN_Size (R= .273, P< .01) and executive age (R= .108, P< .01).

Table 6.7 Correlation Matrix. Pre Financial crisis.

Correlations

LN_TC LN_CC LN_EQC ROA TSR TorbinQ LN_Size Vol Lev Age

LN_TC Pearson Correlation 1 ,486 ** ,643** ,032 ,016 ,009 ,273** -,057 -,038 ,108** Sig. (2-tailed) ,000 ,000 ,295 ,591 ,771 ,000 ,059 ,206 ,000 LN_CC Pearson Correlation ,486 ** 1 ,396** -,037 -,033 -,139** ,279** -,087** ,107** ,171** Sig. (2-tailed) ,000 ,000 ,223 ,271 ,000 ,000 ,004 ,000 ,000 LN_EQC Pearson Correlation ,643 ** ,396** 1 ,061 ,041 ,093* ,167** ,043 -,048 ,106** Sig. (2-tailed) ,000 ,000 ,095 ,264 ,010 ,000 ,232 ,189 ,003 ROA Pearson Correlation ,032 -,037 ,061 1 ,040 ,762 ** -,368** -,202** -,076* -,061* Sig. (2-tailed) ,295 ,223 ,095 ,188 ,000 ,000 ,000 ,012 ,044 TSR Pearson Correlation ,016 -,033 ,041 ,040 1 -,089 ** -,171** ,244** ,024 -,073* Sig. (2-tailed) ,591 ,271 ,264 ,188 ,003 ,000 ,000 ,435 ,016 TorbinQ Pearson Correlation ,009 -,139 ** ,093* ,762** -,089** 1 -,483** -,156** -,136** -,061* Sig. (2-tailed) ,771 ,000 ,010 ,000 ,003 ,000 ,000 ,000 ,043

LN_Size Pearson Correlation ,273 ** ,279** ,167** -,368** -,171** -,483** 1 -,123** -,144** ,063* Sig. (2-tailed) ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,036 Vol Pearson Correlation -,057 -,087 ** ,043 -,202** ,244** -,156** -,123** 1 -,107** -,114** Sig. (2-tailed) ,059 ,004 ,232 ,000 ,000 ,000 ,000 ,000 ,000 Lev Pearson Correlation -,038 ,107 ** -,048 -,076* ,024 -,136** -,144** -,107** 1 ,048 Sig. (2-tailed) ,206 ,000 ,189 ,012 ,435 ,000 ,000 ,000 ,114 Age Pearson Correlation ,108 ** ,171** ,106** -,061* -,073* -,061* ,063* -,114** ,048 1 Sig. (2-tailed) ,000 ,000 ,003 ,044 ,016 ,043 ,036 ,000 ,114

* means 10% significance, ** means 5% significance, *** means 1% significance.

The natural logarithm of cash compensation is also positively significantly correlated with the dummy variables LN_Size (R= .279, P< .01) and executive age (R= .171, P< .01). However, a significant negative relation can be observed between LN_CC and the firm performance variable Tobin’s Q (R= -.139, P< .01). While not significant, the natural logarithm of cash compensation is also negatively correlated with the other firm performance variables TSR and ROA. This was to be expected, due to the declining firm performance and stable base salary. The natural logarithm of equity based compensation only shows a significant correlation with the firm performance variable Tobin’s Q (R= .093, P< .05), and no significant correlation with the other firm performance variables ROA and TSR.

Furthermore, the independent variable Tobin’s Q is significantly negatively correlated to the independent variable TSR (R= -.089, P< .01) and positively correlated with ROA (R= .762, P< .01). Never the less, the correlation between the independent variables does not seem to lead to multi-collinearity.

6.3 Regression results

Table 6.8 presents the regression results.

Independent variable

Independent Variable

LN_TC LN_TC LN_TC LN_CC LN_CC LN_CC LN_EQC LN_EQC LN_EQC

ROA 0.023*** 0.000 0.025*** (5.024) (0.09) (3.305) TSR 0.326*** 0.462*** 0.237 (3.270) (5.120) (1.314) Tobin’s Q 0.095*** -0.11 0.238*** (4.192) (-0.525) (5.877) Financial Crisis 0.0001 -0.184 0.096*** 0.012** -0.424*** 0.055* 0.011 -0.60 0.159*** (0.993) (-1.498) (2.714) (2.315) (-3.822) (1.697) (1.114) (-0.279) (2.655) Vol 0.010*** 0.005** 0.009*** 0.006*** 0.003 0.005*** 0.020*** 0.013*** 0.021*** (4.464) (2.458) (4.343) (2.953) (1.503) (2.702) (5.494) (3.503) (5.838) Lev -0.066 -0.267** -0.003 0.392*** 0.344*** 0.374*** -0.075 -0.294 0.183