CEO cash and non-cash
incentives and their relation
to firm value
Student name: Janneke van der Molen Student number: 1998536
Program: Msc. Accountancy Supervisor: S. Mukherjee Coassessor: V.A. Porumb Date: 20th of June 2016
Word count: 10.530 Abstract:
In 1990 Jensen and Murphy published an article in which they showed how much CEO compensation was aligned with shareholder interests. They estimated the effect that a $1000 change in firm value would have on a CEO’s wealth through the fundamental forms of compensation. They show that stock options are a far larger incentive for a CEO to work in shareholder interests than direct cash based compensation.
I have taken these conclusions and used similar OLS regressions to retest these relationships for the years 1993-2010, in order to determine whether the forms of compensation which have changed the most (salary and bonuses and stock options) have also changed in their magnitude of incentive. I find evidence that supports this and give a yearly overview of estimates of the incentives. Furthermore, I find that the magnitude of the incentives provided by direct cash compensation and the profit gained from stock options are inversely related to firm size.
1
Content
Introduction ...2
1. Theoretical framework ...3
1.1 The relationship between CEO wealth and firm performance ...3
1.2 The year effect on the relationship between CEO wealth and firm performance ...6
1.3 The size effect on the relationship between CEO wealth and firm performance ...7
2. Research methodology ...8
2.1 The models ...8
2.2 Samples and design of the analysis ...9
3. Results ... 10
3.1 Hypothesis 1: The magnitude of the incentives ... 11
3.2 Hypothesis 2: The year effect ... 13
3.2.1 cash based compensation ... 13
3.2.2 awarded stock options ... 14
3.2.2 exercised value of stock options ... 15
3.3 Hypothesis 3: The size effect ... 15
3.3.1 cash based compensation ... 15
3.3.2 awarded stock options ... 16
3.3.3 exercised value of stock options ... 16
4. Discussion ... 19
4.1 Hypothesis 1: The magnitude of the incentives ... 19
4.2 Hypothesis 2: The year effect ... 20
4.3 Hypothesis 3: The size effect ... 22
Conclusion ... 22
Appendix ... 24
2
Introduction
In the past few decades, one of the research areas which have been getting the most attention, is that of the Chief Executive Officer’s (CEO)compensation. Its importance is highlighted by Jensen and Meckling (1976) and later built upon by Fama (1980) and Fama and Jensen (1983), who discuss the concept of the agency problem arising from the separation between ownership and control in a firm. They argue that a CEO’s interests may diverge from those of the shareholders and that it is impossible for the shareholders to monitor all of the CEO’s actions. As a result, top management may take a firm into a direction that does not maximize the value for the owners, or that may possibly even result in a lower firm value (Jensen & Meckling, 1976; Jensen, 1986). By linking CEO compensation to the firm performance or firm value, such an event may be prevented, as harming firm value will then mean decreasing CEO wealth as well.
In 1990 Jensen and Murphy researched how performance pay and incentives relate to each other for the CEO with data from the years1974-1986. The goal of their research was to determine how changes in CEO wealth are related to changes in shareholder wealth through several elements of compensation such as salary and stock ownership. In essence, the two authors compared different compensation elements and determined their seperate effect on CEO wealth. They found that a CEO gains wealth of approximately $3.25, when firm
performance increases with $1000. Their findings have led for the authors to conclude that the CEO incentives are extremely small, and to suggest for political pressure and public
antagonism towards large pay changes to be possible reasons.
A different article by Hall & Liebman (1998) disagrees with this, arguing that Jensen and Murphy’s measure of the incentive may give a misleading image, as firm performance is a very large variable. Even though pay performance sensitivity may seem small, the absolute value of the change of CEO wealth in response to a change in firm performance may still be in the millions for large companies. Hall & Liebman construct pay-performance sensitivities with a variety of measures for the years 1980-1994 and conclude there is a strong relationship between CEO wealth and firm performance and this is supported by other researchers
(Murphy, 1986; Joskow & Rose, 1994; Lilling, 2006).
Another finding in Jensen and Murphy’s article is that there is a difference in how strongly different elements of compensation may motivate a CEO to work in the interests of the shareholders. They estimate the ‘magnitude of incentive’ provided by a form of
compensation, described as the “dollar change in the CEO’s wealth associated with a dollar change in the wealth of the shareholders” (Jensen & Murphy, 1990, p.4). For example, they found that a $1.000 dollar increase in shareholder wealth could be associated with an increase of 2.2 dollar cents of salary and bonus and an increase of 14.5 dollar cents of value of
outstanding stock options. These findings show that salary and bonuses are not the only part of compensation that may motivate a CEO and that a CEO may depend more on a form of non-cash compensation to achieve wealth, than through their cash compensation. Since the findings of this paper, the use of stock options has grown dramatically (Hall & Liebman, 1998; Frydman & Jenter, 2010), and there have been changes in reporting regulations concerning CEO compensation. The pay-performance sensitivity of each type of compensation may differ to twenty years ago.
3
The main research question of this paper is therefore: Have the magnitudes of the incentives provided by salary and bonus and by stock options changed since Jensen and Murphy’s paper in 1990? I have retested the relationship between firm value and changes in CEO wealth through salary and bonus and the issuance and exercise of stock options on the basis of more recent data, specifically from the period of 1993-2010. Moreover, I have tested this
relationship for each year within the sample and analysed the trend that this relationship may show over the years. Lastly, I aim to determine whether the pay-performance sensitivity of both salary and bonus and stock options change for CEOs who work in firms of different sizes.
I find evidence for a positive relationship between changes in past firm performance and both changes in CEO wealth through cash based pay and changes in CEO wealth through stock options. The findings support earlier research in showing that incentives in the form of stock options are substantially larger than cash compensation, as CEOs on average are able to gain a large increase in wealth on exercise of the stock options. Furthermore, the incentives provided by these two types of compensation are not constant over time. Especially the incentive provided by the value which a CEO is able to gain on exercise of stock options has grown dramatically until the year 2000. After 2000 the incentive provided by stock options decrease. I suggest that this is the result of both stock market crashes and the implementation of the Sarbanes-Oxley Act which has made it more difficult to perform backdating. Moreover, I find evidence that the incentives of cash based compensation and the exercised value of stock options decrease with firm size.
This paper adds to previous research by showing that the magnitudes of incentive provided by cash compensation and stock options has changed over the years and that the difference between these two has grown even larger. Furthermore, I support earlier findings that compensation incentives are inversely related to firm size, but add by showing that this applies to both cash and non-cash compensation.
In the upcoming section I describe the theory behind the relationship between changes in CEO wealth and firm performance and develop hypotheses to test. Section 2 contains a description of the research method and expectations. Section 3 describes the results. In section 4, I discuss the results of the hypotheses, interpret them and compare them to Jensen and Murphy’s (1990) earlier findings.
1. Theoretical framework
1.1 The relationship between CEO wealth and firm performance
Much on the literature on CEO compensation is focused on the fact that it has grown a lot over the past years (Bebchuck & Grinstein, 2005, Frydman & Jenter, 2010). However, the composition of CEO compensation has changed over the years as well. Currently,
compensation is mainly comprised of “five basic components: salary, annual bonus, pay-outs from long-term incentive plans, restricted option grants, and restricted stock grants” (Frydman & Jenter, 2010, p.81). In essence, they are forms of cash based pay and non-cash based pay. Two components which have changed the most relative to the other elements of compensation are stock options and salary and bonuses (Frydman & Jenter, 2010). I will discuss these in this paper.
4
In 1970-1979, on which part of Jensen and Murphy’s research is based, stock options made up 11 percent of total compensation, whereas salary and bonus fell around 84 percent. Since then, the relative standing of stock options to other compensation elements has grown greatly, with it reaching its top in the year 2000 of around 49 percent versus 38 percent for salary, bonuses and long-term incentive plans (Frydman & Jenter, 2010). This implies that stock options have played a greater role in a CEO’s wealth than they did during 1974-1986, whereas cash-based compensation seems to have lost on importance as the main determinant of CEO wealth. Several studies support this, by finding that changes in the value of stock options as a result of changes in firm performance are much larger than those in salary and bonuses (Hall & Liebman, 1998; Core et al., 2005) A distinction between salary and bonuses and stock options in determining the relationship between changes in CEO wealth and firm performance is therefore important.
Moreover, a distinction should be made between the awards of the stock options and the exercise of the stock options. Stock options give CEOs the right to buy firm stock in the future at a strike price, also known as the exercise price. Usually the strike price is set at the value of the stock at the grant date (Bianchi, 2015, p.). The value of the awarded stock options represents the value the firm is willing to offer a CEO. For the firm, awarding stock options means that the firm cannot sell this particular batch of stock to other interested buyers, and therefore cannot sell it at the future market price (if it is above strike price). For the CEO, the value of the awarded stock options in a year represents a right, not so much the actual increase in CEO wealth, as the CEO may still decide not to exercise the stock options. Having the chance to exercise options will still increase CEO wealth as well. The actual increase in CEO wealth from stock options is the difference between the market price and the strike price from the stock at exercise, also known as the exercised value of stock options.
I expect the magnitudes of incentives that salary and bonus, the issuance of stock options and the exercise of stock options offer to a CEO to be different from each other, as this was one of the results found in Murphy and Jensen’s paper and others (Hall & Liebamn,1998; Core et al., 2005).
Nyll-hypothesis 1a: The relation between change in firm performance and change in CEO wealth through compensation does not differ between direct cash-based compensation, awarded stock options and the exercised value of stock options.
Alternative hypothesis 1a: The relation between change in firm performance and change in CEO wealth through compensation differs between direct cash-based compensation, awarded stock options and the exercised value of stock options. Performance pay is widely accepted as a method of aligning the interests of a CEO with the shareholder interests. How much effort the CEO will put into these interests, “will depend on the executive’s incentive contract” (Murphy, 1986) and its effect on the CEO’s wealth. A performance-based compensation will only be effective in motivating the CEO to work towards the goals of the shareholders, when taking actions in the shareholders’ interests results in higher CEO wealth than any other actions the CEO may take to increase his own wealth. Therefore, any form of performance-based compensation must be positively linked with shareholder wealth and the compensation must be higher than any possible private benefits which the CEO will enjoy from alternative actions (Hall & Liebman, 1998). This means that the pay-performance sensitivity of compensation does not have to be 100%, as the
5
CEO’s private benefits of an action, which does not maximize shareholder value, are unlikely to be as large as the shareholder benefits of the optimal action. However, it does still imply that the relationship between the change in CEO compensation and the change in firm value is positive. Because of this, my expectation for the magnitude of incentive provided by cash based pay is the following:
Null-hypothesis 1b: There is no relation between change in firm performance and change in CEO wealth through current cash compensation.
Alternative hypothesis 1b: There is a positive relation between change in firm performance and change in CEO wealth through current cash compensation.
The other element of compensation which seems to have dramatically changed since Jensen and Murphy’s findings is the issuance of stock options. Assuming the CEO is able to increase firm value over the years before the exercise date, the CEO is able to buy the stock at a strike price lower than the market price. In essence, the firm is awarding the CEO a chance to create CEO wealth in the future, without risking any current wealth. If a CEO does not manage to influence the firm value such that the market price of stock is above the strike price, the CEO may simply decide to not buy the stock and therefore does not lose any wealth.
Awarding stock options is a strong example of a form of compensation of the agency theory’s implication of aligning the interests of the CEO with those of the shareholders through
compensation. With this form of compensation the firm can both influence future decisions of the CEO and reward the CEO for past behaviour at the same time. It is therefore my
expectation that, should firm performance rise relative to earlier performance, firms will on average reward a CEO by offering more stock options than before. As a result, CEO wealth increases, because he has the chance to make a profit out of these stock options.
What happens to the value of the awarded stock options then? This value is made up from the number of stock awarded, multiplied by the strike price. The CEO would want to receive the right to a larger number of shares, because he has a larger chance to make a profit of them. However, it is in his interest to keep the strike price as low as possible, so that the actual increase of his wealth, the exercised value of the stock options, may be as large as possible. The strike price is usually set at the market price of the stock options at the time of the grant date. This is the lowest the strike price can be set, without incurring expenses for the firm (Bianchi, 2015). If firm value increases, then the market price of stock increases as well, resulting in the increase of the value of awarded stock options.
One way a CEO may influence the strike price, is by spring loading. This is occurs when a CEO uses the information expressed to the outside to influence the share price. The CEO may “release bad news” to the public or “withhold good news” before receiving stock options, so that share price is lower than what it should be (Bianchi, 2015, p. 215). Another method is backdating, where the grant date of an option is manipulated. The grant date is recorded as different from when the stock options are actually rewarded. Instead the board chooses a date where the market price of shares is very low and then records both this date and stock price as the grant date and strike price (Bianchi, 2015). Once again, the result is a strike price lower than that of the current market price of shares.
Using these methods, the CEO will be able to be awarded more stock options with a better firm performance, but also a strike price as low as possible. This implies a positive
6
relationship with better performance. The magnitude of incentive provided by the awards of stock option should be positive.
Null-hypothesis 1c: There is no relation between change in firm performance and change in CEO wealth through the issuance of stock options.
Alternative hypothesis 1c: There is a positive relation between change in firm performance and change in CEO wealth through the issuance of stock options. The third method for firms to increase CEO wealth through compensation is the exercised value of stock options. This is the profit that the CEO manages to make from exercising his options. It is what a CEO is most interested in concerning stock options, assuming his goal is to maximise his wealth.
The exercised value of stock options is highly dependent on the growth of firm value since the grant date. Figure 5 in the appendix shows in a visual manner how market capitalization has developed from 1990-2010. As the figure shows that, overall, market capitalization has grown, I believe that CEOs on average will have been able to achieve market prices higher than the agreed upon strike prices. Moreover, the methods of spring loading and backdating which I discussed earlier, allow for an easy achievement of a lower strike price than market price of stock. Thus, a CEO is more likely to achieve a large and positive exercised value of stock options than not. Therefore, as the market price of stock is strongly related to firm performance of the past years and strike price can be manipulated, I suspect that the exercised value of stock options will have a positive relationship with past firm performance. Therefore, the magnitude of incentive provided by the exercised value of stock options should be
positive.
Null-hypothesis 1d: There is no relation between change in firm performance and change in CEO wealth through the exercise of stock options.
Alternative hypothesis 1d: There is a positive relation between change in firm performance and change in CEO wealth through the exercise of stock options.
1.2 The year effect on the relationship between CEO wealth and firm
performance
As I mentioned earlier, because total compensation and emphasis on certain elements of CEO compensation has changed, it is entirely possible that the incentives given to CEOs to work in the shareholder’s interests have also changed over the past year. Hall and Liebman (1998) support this by finding an increased relationship between changes in CEO compensation and firm performance over the years 1980 and 1994. They suggest that this is because of an increased interest of the board in pay-performance relationships, or because the board sees stock options (which have increased over the past years) as a less noticeable way to pay a CEO more.
Joskow and Rose (1994) found over the period of 1970-1990 that the pay-performance sensitivity of total compensation is not constant over the years. I aim to determine the relationship between changes in firm value and changes of CEO wealth through cash based compensation and stock options for each year for more recent data.
7
An analysis of these results may help determine whether regulations may have had influence on the pay-performance sensitivity for both cash based compensation as well as stock options. For example, the Sarbanes-Oxley Act introduced in 2002 laid the foundation for new
compensation reporting regulations which would make the compensation of CEOs more transparent to the public. One of these regulations was implemented in 2006, which mandated that awarded stock options be reported to the SEC within two business days, with the grant date and the chosen strike price reported as well. Backdating would therefore be much harder to accomplish (Bianchi, 2015). As a result, achieving a high exercised value of stock options is harder to achieve and the magnitude of incentive provided by the exercised value of stock options should be lower from 2006 on.
Moreover, stock market crashes may have had an effect on the changes of CEO wealth
through stock options. As stock market crashes may result in very low market prices for stock, likely to be lower than the strike price, achieving a positive exercised value for stock options is less probable. I expect the magnitude of incentive provided by the exercised value of stock options to be lower during years associated with stock market crashes.
Therefore, because of a changed composition of compensation, regulatory changes and stock market crashes, I expect the magnitude of incentive for each form of compensation to differ over the years.
Null-hypothesis 2: Year effect is not present in the relation between change in firm
performance and change in CEO wealth through current cash compensation, awards of stock options and the exercise of stock options.
Alternative hypotheses 2: Year effect is present in the relation between change in firm performance and change in CEO wealth through current cash compensation, awards of stock options and the exercise of stock options.
1.3 The size effect on the relationship between CEO wealth and firm
performance
In line with agency theory, firms of larger size are more complex to monitor. The agency problem is therefore larger. Aligning the interests of the CEO with those of the shareholders is more necessary than in smaller firms, because a CEO in large firms has to make decisions concerning a very large capital structure. Aside from this, firms of larger sizes are more complex to manage (Gayle and Miller, 2009) and therefore CEOs are more subject to risk. As a result, CEOs should be paid more (Edmans & Gabaix, 2009; Gayle & Miller, 2009).
Frydman and Jenter (2010) add to this, writing that the growth in compensation over the past years has been “steeper in larger firms”.
However, larger firms are more likely to be scrutinized by the public than smaller firms (Jensen & Murphy, 1990). The reason for this close examination is what we call the
managerial power view (Bebchuk & Fried, 2005). This view on managers assumes that CEOs have the power to influence their compensation, as opposed to the idea that objective boards are completely responsible for this process. As the CEOs tend to be selfish and wish to maximize their wealth, they use this power to extract rent from the firm, by setting excessive pay (Bebchuk & Fried, 2005). CEOs of large firms are seen as more likely to be awarded excessive pay and therefore their compensation is more closely examined by the public.
8
Jensen and Murphy (1990) find there to be a difference in pay-performance sensitivity for cash based compensation. They determine that the incentives decline strongly in firm size, which is supported by Schaefer (1998) and Baker and Hall (1998).
Bebchuk and Fried also suggest that rent extraction is used more in forms of compensation which are less visible than others, such as stock options (Frydman & Jenter, 2010). As the compositions of CEO compensation has changed so much and stock options have become such a large part of total compensation, I expect the public to scrutinize stock options as well. The pay-performance sensitivity for awarded stock options and the exercised stock options should therefore decrease in size as well.
Thus, I expect the magnitude of incentive for each form of compensation to decrease in size. Null-hypothesis 2: Size effect is not present in the relation between change in firm
performance and change in CEO wealth through current cash compensation, the awards of stock options and the exercise of stock options.
Alternative hypotheses 2: Size effect is present in the relation between change in firm performance and change in CEO wealth through current cash compensation, the awards of stock options and the exercise of stock options.
2. Research methodology
2.1 The models
The goal of my thesis is to determine the magnitude of incentive cash based compensation and stock options give to a CEO. In essence, I aim to determine the relationship between changes in the two forms of compensation and changes in shareholders wealth with a multiple regression analysis. To test this, I use a model introduced by Jensen and Murphy (1990) to determine the relationship between cash based compensation and shareholder wealth (model 1). I add to this by using the same composition for models for the value of awarded stock options and the exercised value of stock options . The models are the following:
Model 1: Δ(salary + bonus) t= α + β1 Δshareholder wealtht + β2 Δshareholder wealth t-1
Model 2: Δ(value of awarded stock options)t = α + β3 Δshareholder wealtht + β4
Δshareholder wealth t-1
Model 3: exercised value of stock optionst = α + β5 Δshareholder wealtht + β6
Δshareholder wealth t-1
As independent variables, I use the change in shareholder wealth from years t and t-1, measured by the change in market value of the firm for each of the models. The sum of the estimates of the independent variables form the magnitude of incentive of that compensation mechanism, b. Like Jensen and Murphy I define b, as “the dollar change in the CEO wealth associated with a dollar change in the wealth of the shareholders” (Jensen & Murphy, 1990, p.4 ). A higher b therefore denotes a better alignment of CEO and shareholder interests. To determine the effect of cash based compensation on the CEO’s wealth, I use model 1, with the change in salary and bonus in year t as the dependent variable. Within bonuses, I include any cash-based non-equity incentive plans which are reported from 2006-2010. There are
9
other forms of cash compensation which can be seen as value increasing incentives, such as short-term incentive plans, but in the interest of comparing the results with Jensen and Murphy, I limit the variables to salary revisions and bonuses.
To determine the magnitude of the incentive that the firms are willing to give in the form of non-cash compensation, I use model 2, with the change in value of awarded stock options in year t as the dependent variable. I measure this as the grant date fair value of options awarded during the year as measured by the firms, as these represent the incentive a firm is willing to give to the CEO for the change in performance they achieve.
The grant date fair value of options does not indicate the true increase of a CEO’s wealth, however. The increase of wealth of the CEO enjoys due to stock options is the difference between the strike price of a stock option and the market price of the stock on the exercise date. This is also known as the exercised value of stock options. I use this as the dependent variable for model 3, to determine the effect of non-cash compensation on the CEO’s wealth.
2.2 Samples and design of the analysis
I have collected compensation and firm value data from Standard & Poors’ Execucomp database from firms in the United States in 1992-2010 and have paired this with data on firm value from CRSP. Using this data, I constructed a sample for each model. Model 2 is
restricted in the sense that only data on the fair value of stock options was collected from 2006 on. The reason for this is that, before the mandated compensation disclosure regulations from the Securities and Exchange Commission in 2006 (www.sec.gov), disclosure of this variable was optional. Considering differences over the years, this results in data samples from 1993-2010 for the change in cash compensation and the exercised value of stock options and from 2007-2010 for the change in fair value of stock options.
Furthermore, the regulations on disclosure of compensation have changed in 2006, which have an impact on the reported salaries and bonuses. Before 2006, plan based awards in the form of cash could be recorded under bonuses (www.fasb.org). After the compensation disclosure regulations mandated by the SEC, such awards must now be reported under “non-equity incentive plans”. Non-“non-equity incentive plans are, as defined by the sec in their final rule, “an incentive plan or portion of an incentive plan […] under which awards are granted that [do not] fall within the scope of FAS note 123” (www.fasb.org). As FAS note 123 considers share-based payment, all incentive plan awards in the form of cash payment now fall under non-equity incentive plans and are no longer reported as bonuses. From 2006 on, I have included non-equity incentive plans in the cash based compensation which consisted of salary and bonuses before 2006.
I have removed firm-year combinations for which not all data for the variables in that model was present and adjusted outliers for each variable, because they would otherwise have an unreasonably large effect which could bias the results. I have done this by determining an area of the size of three standard deviations around the mean, which would encompass acceptable data values. Data which lie beyond this area, are set at the value of the border of this area. The initial sample consisted of 2.930 U.S. firms, 6.630 CEOs and 40.722 firm-year
combinations. After removing firm-year combinations for which data was not present, I was left with the following samples for each model: Sample A for model 1 (cash based
10
compensation) consisted of 2.681 firms, 5.059 CEOs and 25.929 firm-year combinations for cash compensation; sample B for model 2 (awarded stock options) consisted of 1.746 firms, 2.151 CEOs and 6.028 firm-year combinations for the change in value of stock compensation; sample C for model 3 (exercised value of stock options) consisted of 2.956 firms, 2.699 CEOs and 23.639 firm-year combinations.
For testing hypothesis 1, I used these three main samples. For testing the year effect in hypothesis 2, I used the same observations, only split them into smaller subsamples for each year. For cash based compensation, this means that I regressed model 1 on sample A, but only on the observations from 1993. Then I once again tested the model, but only then for 1994. I did the same for the years 1995-2010. This resulted in estimates for β1 and β2 for each year.
For the change in CEO wealth from awarded stock options I regressed model 2 on sample B, but first only for the observations from 2007. I tested this also for 2008, 2009 and 2010. For the exercised value of stock options I used the same method, regressing model 3 on sample C for each year in 1993-2010.
For hypothesis 3, I partitioned the main samples A, B and C into subsamples for each size class, like I did for hypothesis 2 for each year, and tested each subsample for β1 and β2. This
left me three subsamples for each model: a subsample containing information from sample A, but only with observations from large firms, then one subsample from sample A for the medium sized firms and one subsample from sample A for the smaller firms. I categorized firms into a certain size classification by determining which firms would make up the top 30%, the middle 40% and the bottom 30% of the total market valuation of each year. The top is size class one, the middle size class two and the firms with the 30% lowest market
valuations are size class three.
An initial analysis for multicollinearity was performed by assessing the correlation
coefficients and the VIF values. After this, I tested the hypotheses for each model, using the OLS regression method for two-tailed testing. For each hypothesis, I tested models 1, 2 and 3 with only the change in shareholder wealth of the current year (step 1), before adding the change in shareholder wealth of the previous year to the model (step 2). I used the estimates of the most appropriate regression to determine whether I could accept the hypothesis or not. In this case, that would be the regressions results from step 2.
3. Results
Here, I will discuss the results of the analysis. The descriptive statistics of the samples of each model can be found in tables 1a-c. Note that all amounts are in $1000. Additionally, the associated correlations of each model can be found in tables 2a-c. The regression results for each of the hypotheses can be found in table 3 until table 9b. Each of these tables are included in the appendix.
A review of the correlation tables shows no absolute values larger than 0,7, signifying no strong correlations between variables in any of the three models. Moreover, no VIF values higher than 10 have been found in any of the tables with the regression results. I therefore conclude that there is no multicollinearity in the models I designed.
I will discuss the results for each hypothesis in the following three chapters. First, I will give you an explanation for the included tables with the regression results. In table 3, you can find
11
the regression results for hypothesis 1, where I estimate the relationship between change in firm value and the three forms of compensation. All three models consist of the following function
independent variable = α + β1 Δshareholder wealtht + β2 Δshareholder wealth t-1
α represent the intercept, the change in compensation when firm value stays the same. β1 and
β2 represent the estimates for the relationship between the change compensation and the
change in firm value. Every time I did the regression, I performed it in two step. In the first step (1), I only included the change in shareholder wealth of the current year. I only estimated β1. In the second step (2), I also included the change in shareholder wealth of the previous
year and thus estimated β1 and β2. In table 3 I put these two steps in one table. From
hypothesis 2 on, these two steps are separated in two tables (usually table a and table b). For example, table 4a gives regression results for cash compensation for each year, but here I only include the change in shareholder wealth for the current year. In table 4b the second
independent variable is included in the regression for cash based compensation. For easy reference I have put clear titles above each table.
The results in tables 3 until 9b are all in $1000 dollars. This means that, should firm value not increase, cash based compensation would increase with $30.481. Should the firm value of the current year change, for each increase of $1000 in firm value, CEO cash compensation would increase with 5,9 dollar cents. The total magnitude of incentive for a form of compensation is b, which is the sum of the estimates of β1 and β2. It represents the change in compensation for
each $1000 dollar change in firm value of the past two years.
The adjusted R-squared, R-squared change F-value and VIF help determine whether the model is valid and can be used
3.1 Hypothesis 1: The magnitude of the incentives
The regression results for hypotheses 1 can be found in table 3. Regression of the change in salary and bonus with shareholder wealth (model 1) results in an adjusted R-squared of 0,048, signifying shareholder wealth of the current year explains around 4,8% of the variance in the change of salary and bonus. Adding the change in shareholder wealth of the previous year into the model increases the R-squared very slightly, still leaving it around 0,048. The F-value of the model is significant at the 1% level, meaning the joint effect of the change in
shareholder wealth of the current and the previous year is statistically significant. The regression of the change in the value of stock options (model 2) implies an explanation of 0,1% of the variance in the change of the value of non-cash compensation. Including the change in shareholder wealth of the previous year adds another 0,3%. Here, the two
independent variables are also highly significant at the 1% level. Model 3 relates the change in shareholder wealth with the exercised value of stock options and explains around 1,6% of the variance, 4,6% if the change in shareholder wealth of the previous year is included. The model itself is significant at the 1% level.
As including the change in shareholder wealth of the previous year in all three models leaves us with statistically significant regressions for all three forms of compensation, I will continue with the estimates of step 2 for each form of compensation.
12
As can be seen in table 3 in the appendix, the intercepts of all three models are statistically significant. The intercepts of the change in cash compensation and the exercised value of stock options are significant at a 1% level and the intercept of the change the value of stock options is statistically significant at the 10% level. The change of the shareholder wealth in the current year is highly significant for the change in cash compensation and the exercised value of stock options, while it is significant at the 10% level for the change in value of stock options. The change of shareholder wealth in the previous year is highly significant for the change in value of stock options and the exercised value of stock options and statistically significant at the 5 percent level for cash compensation..
Figure 1 below shows the results of table 3 in a visual manner. It represents the calculated b, the total magnitude of incentive of the compensation. It is the sum of the estimates of both independent variables. Figure 2 represents the change in compensation when a CEO is not able to change firm value in comparison to the previous year (the intercept). Note that the results in the tables are in thousands of dollars, whereas the figures show the results in a more convenient manner: in dollar cents for b and in dollars for the intercept.
The interpretation of the results in the figures is as follows. When a CEO does not increase the shareholder wealth in comparison to last year, compensation in the form cash increases with $30.481 as opposed to last year. The fair value of awarded stock options decreases with $63.832 in comparison to the awarded options of last year and the CEO is able to achieve an exercised value of stock options of $1.684.731. For each increase of shareholder wealth of $1000 dollar, cash compensation increases with 6,28 dollar cents, the value of stock options increases with 1,11 dollar cents and the CEO is able to achieve an exercised value of stock options of 37,58 dollar cents. Remarkable is that the intercept for awarded stock options is negative.
Within hypothesis 1a I proposed that the magnitude of incentive would differ for each form of compensation. The results support this, showing that the magnitude of incentive from the exercised value of stock options is much higher than that of cash compensation or the value of awarded stock options. I therefore reject null-hypothesis 1a and have found some evidence for the alternative hypothesis.
As the magnitudes of incentives are positive and significant for all three models, the null-hypotheses 1b-d can be rejected and null-hypotheses 1b-d are supported by the results. Here, I suggested that the three forms of compensation would be positively associated with changes in firm value.
13
3.2 Hypothesis 2: The year effect
The results for hypothesis 2 can be found in tables 4a-6b. Table 4a shows the results for the regression of the change in cash compensation on the change in shareholder wealth for the current year (step 1 of the regression). Table 4b includes the change in shareholder wealth for the previous year in this as well (step 2 of the regression). The same can be said for tables 5a and 5b for the change in the value of the awarded stock options and tables 6a and 6b for the exercised value of the stock options.
3.2.1 cash based compensation
The regression results for cash compensation for hypothesis 2 can be found in tables 4a and 4b. Regression of the change in salary and bonus with shareholder wealth (model 1) results in an adjusted R-squared between 0,008 and 0,089 in the regressions made for the years 1992-2010. Adding the change in shareholder wealth of the previous year into the model increases the R-squared very slightly for all years. The F-value of the model is significant at the 1% level for all years in step 2. The only exception to these observations is the year 2009, where the joint effect of the independent variables is not significant at all in step 2. Instead the regression with only the change in shareholder wealth for the current year is significant at the 10% level.
Figure 1: incentive b - change in compensation when shareholder wealth
increases with $1000
Figure 2: The change in CEO wealth insensitive to past performance
14
As step 2 is both statistically significant and adds to R-squared in all years except 2009, I use it as the appropriate regression for my hypothesis. In 2009 I will use step 1 as the appropriate regression.
For an overview of the intercept and the estimators β1 and β2 separately,I refer you to tables
4a and 4b. The estimates for change in shareholder wealth in the current year are statistically significant for all years. The estimates for change in shareholder wealth in the previous year is significant for all years except 1994, 1995, 2000, 2001 and 2005. The intercept of the model is significant for all years except 1993, 2001, 2003, 2004 and 2008.
For a visual overview of the estimated b’s for cash based compensation I refer to figure 3. It shows in a visual way what the effect of a $1000 change in past firm performance has on CEO cash based compensation in dollar cents. Just by taking a look at the cash-based part of the graph, one can conclude the alignment between CEO interests and shareholder interest through cash based compensation differs in years, even if we only take a look at the years the estimates are significant. You can see in figure 3 that cash compensation was more aligned with shareholder interest during 1993-1995 than most of the years after that. In fact, the incentive provided by cash compensation decreases until the year 2000, where it starts to rise again. Another point which should be made, is that the magnitude of the incentive of cash compensation seems to have grown more turbulent after 2006. Therefore, the null-hypothesis 2a is rejected and evidence has been found for alternative hypothesis 2a.
3.2.2 awarded stock options
The regression results for the change in value of the awarded stock options for hypothesis 2 can be found in tables 5a and 5b. The regressions of the change in the value of stock options (model 2) over the years 2007-2010 come with R-squared between 0,002 and 0,001 and the R-squared change for adding change in shareholder value of the previous year is positive for all years. The F-value for the complete models are significant at the 1 percent level for year 2009, significant at the 10% level for 2007 and 2010 and not statistically significant for 2008. As step 2 of the regression adds to R-squared for all years, I continue working with the
regression including the change in shareholder wealth for the previous year. It should be noted that the regression for 2008 is not statistically significant and can therefore not be completely reliably assumed. However, the results for 2008 are still economically significant.
For an overview of the estimators β1 and β2 separately,I refer to table 4b. And like cash based
compensation, the intercept for awarded stock options can be found in figure 4 (as well as table 4b). It is only in 2008 that the intercept of awarded stock options is positive. Please note that no intercept is statistically significant, nor are the change in shareholder wealth in the current year for 2007 and 2008 and the change in shareholder wealth in the previous year for 2007 and 2009. Further interpretation of this result can be found in the discussion section. For an overview of the estimated b’s for awarded stock, please once again take a look at figure 3. Like cash based compensation, a look at the graph shows that the incentive a firm offers to a CEO in the form of stock options, is not constant over the years, but fluctuates. In fact, the incentive of awarded stock option are positive in 2007 and 2010, but in 2008 and 2009 it is negative. Therefore null-hypothesis 2b can be rejected and evidence is found for the alternative hypothesis 2b.
15 3.2.2 exercised value of stock options
The regression results for the exercised value of stock options for hypothesis 2 can be found in tables 6a and 6b. Model 3 relates the change in shareholder wealth with the exercised value of stock options for the years 1993-2010 and explains between 1,2% and 25,2% of the
variance in the exercised value of stock options. Including the change in shareholder wealth for the previous year adds in R-squared for all years. Within all regressions, the joint effect of the change in shareholder value of the current and the previous year is statistically significant at the 1% level. Therefore, step 2 is the appropriate regression for this relation and the change in shareholder wealth of the previous year should be included. I will continue analysing the estimates for step two in table 6b.
In table 4b you can find the results for the estimators β1 and β2, as well as the intercept.
Furthermore, the intercept is presented in a visual manner in figure 3 and b, the magnitude of the incentive for the exercised value of stock options is represented in figure 4. All
independent variables and all intercepts of each year are statistically significant. The only exceptions to this are the estimates for the change in shareholder wealth in the current year for 1993 and 2001.
Figure 3 shows what the effect of a $1000 change in past firm performance has on the exercised value of stock options over the past years. As expected, the magnitude of the incentive of the exercised value of stock options grows on average until the year 2000, where it reaches its peak, before it decreases over the years. Notable is that it takes a dramatic nosedive during the years 2001 and 2002 (where it is even negative!), before recovering to the same level as around the year 1997. Also, during the years 2008 and 2009 it decreases again and in the year 2010 it takes another nosedive.
Like cash compensation and the awarded stock options, the null-hypothesis that there is no year-effect in the incentive of the exercised value of stock options can be rejected, as there is clearly a large fluctuation of b over the years. Evidence has been found for alternative hypothesis 2c.
3.3 Hypothesis 3: The size effect
3.3.1 cash based compensation
Tables 7a and 7b shows the results from regressions on cash based compensation for large firms, medium firms and small firms.
The regression for large firms has an adjusted R-squared of 0,05, which is increased with 0,0027 by adding the change in shareholder wealth of the previous year to the model. The regression for medium (small) firms has an adjusted R-squared of 0,071 (0,062), which increases with the addition of the second independent variable. All models of the regression in step 2 are significant at the 1% level. Therefore, step 2 is the most appropriate regression for all firms and I will continue with the estimates in table 7b. All intercepts and estimates are significant in these three regressions, except for the change in shareholder wealth of the previous years for the large firms.
The incentive provided by cash based compensation is 4,3 dollar cents for large firms, 6,7 dollar cents for medium sized firms and 15,1 dollar cents for small sized firms. This is also visually represented in figure 5, together with the other forms of compensation. As all of these
16
estimators are statistically or economically significant and the incentive from large firms is smaller than those from small firms, the null-hypothesis 3a can be rejected and evidence has been found for the alternative hypothesis 3a.
3.3.2 awarded stock options
Tables 8a and 8b concern the regressions done on awarded stock options for differently sized firms.
The regression done for large firms comes with a negative squared of 0,010. Though R-squared does increase with the addition of the second independent variable, the model has a statistically insignificant F-value (for both step 1 and step 2) and both the estimates and the intercept are insignificant. A possible reason for these results is the small sample of only 79 observations.
The models for medium (small) firms come with an R-squared of 0,006 (0,002) which does increase after adding the second variable. Both models have a statistical significant joint effect in step 2 of the regression and both intercepts are not statistically significant. The estimates of β1 and β2 are statistically significant, though.
A visual representation of the results can be found in figure 5. Large firms offer an incentive tied to stock awards of -1,60 dollar cents for each increase in firm value of a $1000 dollars. For medium sized firms this is 2,29 dollar cents and for small sized firms 0,29 dollar cents. The incentive is not constant over firm size for stock option awards and the null-hypothesis 3b, which assumes there is no size effect, can be rejected. There is a size effect present in the magnitude of incentive provided by awarded stock options. However, the incentive does not decrease with firm size, which does not support the expectation. Instead, the incentive seems to be largest for medium sized firms, negative for large firms and positive for small firms.
3.3.3 exercised value of stock options
Tables 9a and 9b contain the results from the regressions done for the exercised value of stock options. The model for large (medium/small) firm seems to explain 1,0% (1,6% / 0,1%) for the variance in the exercised value of stock options. Adding the change of shareholder wealth of the previous year enhances this with 4,9% (2,7% / 2,3%). All three models are significant at the 1% level. Therefore, including the change in shareholder value for the previous year is appropriate and I will continue to use the estimates from 9b. All intercepts and estimates are statistically significant
A look at figure 5 shows that the incentive provided by the exercised value of stock option decreases when a firm grows larger. CEOs of small firms manage to achieve 45,23 dollar cents for each $1000 increase of firm performance. For medium size firms this is 30,25 and for large firms it is 20,05.
Because evidence points towards a decreasing incentive, the null-hypothesis 3c that size does not matter has to be rejected. The alternative hypothesis 3c is supported.
17
18
19
4. Discussion
4.1 Hypothesis 1: The magnitude of the incentives
Within alternative hypothesis 1a, I expected the magnitudes of incentives for the three forms of compensation to differ from each other. The results I have found support this. However, more reliable and conclusive research should be done into this area. Within alternative hypotheses 1b-d I expected cash compensation, stock options awards and the exercised value of stock options to be positively related to past firm performance. The results support these hypotheses and have led me to reject the null-hypotheses. Furthermore, I expected that the incentives provided by the two forms of compensation are larger than Jensen and Murphy’s findings.
For each increase of shareholder wealth of $1000, the CEO is rewarded with 6,3 dollar cents in cash and 37,6 dollar cents in the exercised value of stock options. The firms seems to reward positive past performances by offering the CEO the chance to buy 1,1 dollar cents of stock options in the future for each increase of $1000 in firm value in the past, which gives him the chance to create more CEO wealth on exercise. Jensen and Murphy (1990) found cash based incentives of 2,19 dollar cents per $1000 increase of shareholder wealth, as well as 14,5 dollar cents from the exercised value of stock options. Cash based incentive seems to have almost tripled and the incentive provided by stock options has more than doubled! This implies that pay-performance sensitivity has gotten a larger role since Jensen and Murphy’s research.
Figure 5: incentive b - change in compensation when shareholder wealth increases with $1000
for firms of different sizes
Figure 6: the change in CEO wealth insensitive to firm performance for large,
20
Furthermore, the implication that stock options are more of an incentive for CEOs to work towards shareholder goals than direct cash compensation is supported here as well. Not only is the pay-performance sensitivity higher than cash compensation, the change in CEO wealth when firm value does not change is over $1.600.000 (see figure 2). Stock options seem to be more important to a CEOs wealth than salary and bonuses.
Another surprising finding is that, on average, firms award their CEOs $63.832 less than the previous year when they have not increased firm value. That a CEO agrees to be
disadvantaged while they have not harmed firm value seems illogical. A possible explanation for that a CEO would agree to this, is when he believes that he is able to ‘outperform’ this disadvantage. He believes he can create enough firm value by the time of the exercise date, so that he achieves an exercised value of the stock options which is larger than the perceived lowered chance to create wealth. Therefore, CEOs must be very narcissistic to agree to a similar incentive. According to Olsen and Stekelberg (2016) “narcissism is a stable, multifaceted personality trait consisting of grandiosity, importance, and inflated self-views”. Narcissists believe in themselves, their capabilities and their entitlement and are more likely to take risks (Chatterjee & Hambrick, 2007; Foster et al., 2011).On average, it appears that their narcissistic opinion comes true, as the CEO is able to achieve very high exercised value of stock options.
4.2 Hypothesis 2: The year effect
In line with hypothesis 2 I expected the incentives provided by cash based compensation and stock options to show fluctuations. The results support this expectation and allow me to reject the null-hypotheses that the incentives are constant over time. Furthermore, I expected cash compensation to show some growth in incentive and stock options to show growth until at least the year 2000, followed by a decline. Aside from this, I expected the incentive of stock options to be susceptible to stock market crashes.
As expected, the incentive provided by the exercised value of stock options has increased from 1993 until 2000. In 2000, 2001 and 2008 the incentives had decreased extremely, which I suspect to be because of bad state of the stock market during those years (see figure 7). As the exercised value of stock options depends heavily on the market price of the stock being higher than the earlier agreed upon strike prices, unexpected stock market crashes can result in a decrease of the market price of stock. The relationship between changes of firm
performance and the exercised value of stock options may even turn negative when firm performance goes down, but the firm awards more stock options to the CEO or reprices stock options in holding with the idea of tying the CEO’s wealth even further to the firm value, so that he is more motivated to increase firm performance. Another possibility is that CEOs performed backdating to earn an exercised value of stock options, even though firm performance and stock price have gone down.
Backdating occurs when the strike price of stock options is manipulated post grant date in such a way that it is set on an especially low level and the CEO may gain even more exercised value than when he buys the stock at the strike price it should have been. In the 1990s
backdating was a widely used practice (Heron & Lie, 2007). The reason for this is found in the managerial power perspective (Bebchuk & Fried, 2005), which assumes that executive compensation is not set by an independent board, but that the CEO has considerable influence
21
in determining his compensation. In doing so, the CEO tries to get as much value from the firm as possible, possibly trying to achieve this in ways which are less obvious. Stock options allowed for this with the possibility of backdating.
With the introduction of the Sarbanes-Oxley Act in 2002, backdating has become more difficult, and almost impossible since the implementation of it by the SEC in 2006 through FAS no. 123 (Bianchi, 2015). Firms are now required to inform the SEC of stock option awards within two business days, as well as the chosen strike price for these options. The possibility of an effect on CEO wealth by backdating may well be an important explanation for the growth of the incentive of the exercised value of stock options until the year 2000. After 2000 and 2001 the incentive of stock options shows a trend of decline. The rules, which made backdating nearly impossible for CEOs and were introduced in 2002, may have resulted in a lower exercised value of stock options and also be the reason for the decline of stock popularity since 2002 that Frydman and Jenter (2010) found. Another reason may be that the stock market crash of 2001 showed CEOs how heavily dependent of firm value they have become and they have started to look for a different manner of compensation to focus on.
Cash compensation was more aligned with shareholder interest during 1993-1995 than most of the years after that. In fact, the incentive provided by cash compensation decreases until the year 2000, where it starts to rise again. This coincides with the introduction of the Sarbanes-Oxley Act, which mandated that compensation should be reported in a more transparent manner to outsiders. In 2006, when these regulations were implemented, the incentive of cash compensation is more turbulent. Remarkable is that the increase of the magnitude of incentive for cash compensation coincides with the decrease of the magnitude of incentive of stock options. As the magnitude of incentive is estimated for only a few years after this moment, more research with data from more recent yours should be done to determine whether these trends are related to each other.
Another point which should be made, is that the magnitude of the incentive of cash
compensation seems to have grown more turbulent after 2006. This could be due to the new compensation regulations in 2006. Another possible reason is the stock market crash in 2008 and the fact that the relationship between cash compensation and firm value is estimated until only 2010. Like earlier mentioned, more research with more recent data is necessary to determine the reason for this observation.
The last unusual observation is the negative relationship between the change in value of awarded stock options in the years 2008 and 2009, which is contrary to the expectation. One conceivable reason is the appearance of repricing stock options during the stock market crashes. Stock repricing occurs when the firm cancels earlier awarded stock options because their strike price is higher than the current market price of stock. Subsequently, the firm awards new stock options with the current market price as the new strike price. This way the firm motivates the CEO to work more in the shareholders’ interests and to try to raise firm performance again. Even though firm value falls, stock options are awarded.
22
4.3 Hypothesis 3: The size effect
According to hypothesis 3 I predicted that the incentive of both cash compensation and stock options would be inversely related to firm size, because larger firms are more closely
examined by outsiders. I find that the magnitude of incentives provided by cash compensation and the exercised value of stock options is indeed inversely related to firm size.
Jensen and Murphy (1990) estimated the same relationship for cash compensation, only dividing the firms into two sizes: the top 50 % and the bottom 50%. They found that the incentive of cash compensation was around 2 dollar cents for large firms and 4,1 dollar cents for small firms. I expected that, should I divide the groups in a more extreme manner, such as focusing on the top 30% and the bottom 30%, that this difference would be stronger. This expectation is realized in the sense that cash compensation for a small firm gives between two to over three times more incentive. I find that the incentive provided by cash based
compensation is 4,3 dollar cents for large firms, 6,7 dollar cents for medium sized firms and 15,1 dollar cents for small sized firms.
Furthermore, I find that CEOs of small firms manage to achieve 45,2 dollar cents of exercised value of stock options for each $1000 increase of firm performance. For medium size firms this is 30,4 dollar cents and for large firms it is 20,1 dollar cents. I am unable to compare this with Jensen and Murphy, as they assumed non-cash incentives would not change over size, since in their opinion it was not subject to political pressure. My findings do imply that a size effect is present in the exercised value of stock options.
For the change in value of the awarded stock options, I do find a size effect, but not one inversely related to firm size. CEOs of large firms are provided with an incentive of -1,6 dollar cents, CEOs of medium firms with 2,29 dollar cents and CEOs in smaller firms with 0,3 dollar cents. More research is needed to be able to explain this curious observation.
Conclusion
Have the magnitudes of the incentives provided by salary and bonus and by stock options changed since Jensen and Murphy’s paper in 1990? I conclude that they have. Not only has the incentive provided by cash compensation nearly tripled, the incentive provided by stock options in the form of their exercised value has doubled as well. Jensen and Murphy’s
implication that non-cash compensation is a higher incentive to CEOs than cash compensation is supported by my findings. Cash compensation in the form of salary, bonuses and non-equity incentive plans provides 6,3 dollar cents to the CEO’s wealth for each increase of firm value with $1000, while stock options provide 37,6 dollar cents. At the same time, the firm is willing to provide the CEO with the chance to buy 1,1 dollar cents more of stock options, to use in creating CEO wealth through exercise of the options.
I conclude that the magnitude of these incentives has changed over the years on which I collected data, namely 1993-2010. The incentive of cash compensation has decreased before the year 2000, after which it started growing. The average incentive of the exercised value of stock options has increased until the year 2000 as well, before starting a descent. Furthermore, the relationship between the exercised value of stock options and firm value is much lower during stock market crashes, as can be expected. The introduction of the Sarbanese-Oxley Act
23
in 2002 and the implementation of more transparent compensation reporting regulations seem to have had an effect on the magnitude of the incentives of the compensation.
Lastly, I conclude that the magnitude of the incentives provided by cash compensation and stock options is inversely related to firm size.
I have added to previous research by retesting the findings of Jensen and Murphy (1990) in a more recent setting. Furthermore, I add to their findings concerning size differences in the magnitude of incentives by estimating the incentives for small, medium and large firms. Lastly, to my knowledge, there has not been an overview of the relationship between changes in firm value and changes in cash compensation and the exercised value of stock options created for recent years.
Future research can be done on the relationship between the value of awarded stock options and firm value for more and more recent years. The current paper is restricted in the sense that only data was collected until 2010 and no data was reported before 2006. This results in not only a short period in which the value of the awarded stock options could be estimated, but also one which is heavily influenced by the stock market crash in 2008.
Furthermore, more can be done concerning the effect of the Sarbanes-Oxley Act and the compensation regulations in 2006 on the magnitude of the incentives provided by cash compensation and the exercised value of stock compensation. The findings imply that the cash compensation has gone up, while stock options have started to play a smaller role in compensation practices after 2006. However, due to the short period and the condition of the stock market during 2008, this data may be biased. More research is needed to determine the actual effect on the incentives and whether the narcissistic opinion of the CEOs on stock options and their capabilities on creating profit from them are still true after 2006.
24
Appendix
Table 1a: Descriptive Statistics - Model 1
N Minimum Maximum Mean Std. Deviation
Δ(salary + bonus) 25.929 -7.379,487 7.393,404 57.502 1.078,593
Δshareholder wealtht 25.929 -23.599.286,824 24.545.482,795 434.261,457 3.997.851,937
Δshareholder wealtht-1 25.929 -22.269.866,537 23.078.661,736 364.186,662 3.615.608,769
Valid N (listwise) 25.929
Table 1b: Descriptive Statistics - Model 2
N Minimum Maximum Mean Std. Deviation
Δ(value of stock options) 6.028 -71.410,100 35.209,400 -66,663 2.693,658
Δshareholder wealtht 6.028 -204.177.726,700 114.985.698,174 -170.378,376 8.482.541,385
Δshareholder wealtht-1 6.028 -204.177.726,700 114.985.698,174 -222.415,611 8.579.694,773
Valid N (listwise) 6.028
Table 1c: Descriptive Statistics - Model 3
N Minimum Maximum Mean Std. Deviation
Exercised value of stock
options 23.639 -99,883 36.680,330 1.768,869 5.223,653
Δshareholder wealtht 23.639 -24.773.555,167 25.361.846,128 276.427,452 4.236.195,821
Δshareholder wealtht-1 23.639 -24.536.002,241 24.895.408,482 186.998,244 4.113.475,124
Valid N (listwise) 23.639
Table 2a: Correlations - Model 1
Δ (s al ar y + bo nu s) Δ sh ar eh ol de r w ea lt h t Δ sh ar eh ol de r w ea lt h t-1 Δ(salary + bonus) 1,000 0,218 *** 0,008 Δshareholder wealtht 0,218 *** 1,000 -0,020 *** Δshareholder wealtht-1 0,008 -0,020 *** 1,000
25
Table 2b: Correlations - Model 2
Δ(v al ue o f st oc k op ti on s) Δ sh ar eh ol de r w ea lt h t Δ sh ar eh ol de r w ea lt h t-1
Δ(value of stock options) 1,000 -0,033 *** 0,062 ***
Δshareholder wealtht -0,033 *** 1,000 -0,178 ***
Δshareholder wealtht-1 0,062 *** -0,178 *** 1,000
***. Correlation is significant on 1% level (two-tailed).
Table 2c: Correlations - Model 3
Exe rc is ed v al ue of s to ck o pt io ns Δ sh ar eh ol de r w ea lth t Δ sh ar eh ol de r w ea lth t-1
Exercised value of stock options 1,000 ,125 *** ,173 ***
Δshareholder wealtht ,125 *** 1,000 -0,004
Δshareholder wealtht-1 ,173 *** -,004 1,000
***. Correlation is significant on 1% level (two-tailed).
Table 3: Regression results hypothesis 1 – All models
Salary + Bonus Stock Options Stock Options Exercised Value
(1) (2) (1) (2) (1) (2) Intercept 31,915 *** 30,481 *** -68,429 ** -63,832 * 1726,273 *** 1684,731 *** Δshareholder Wealth t 0,0000589 *** 0,0000590 *** -0,0000104 ** -0,0000071 * 0,0001541 *** 0,0001550 *** Δshareholder Wealth t-1 - 0,0000039 ** - 0,0000182 *** - 0,0002208 *** Adjusted R-squared 0,048 0,048 0,001 0,004 0,016 0,046 R-squared change 0,0002 0,0032 0,0302 F-value 1298,468 *** 651,594 *** 6,425 13,018 *** 374,986 *** 567,759 *** Highest VIF 1,000 1,000 1,000 1,033 1,000 1,000 b 0,000059 0,000063 -0,000010 0,000011 0,000154 0,000376 n 25929 25929 6028 6028 23639 23639
26
Table 4a: Regression results hypothesis 2 – Model 1 Δ(salary + bonus) - step 1
1993 1994 1995 1996 1997 1998 1999 2000 2001 (1) (1) (1) (1) (1) (1) (1) (1) (1) Intercept 66,079 *** 78,252 *** 25,806 ** 62,837 *** 65,678 *** 15,317 79,325 *** 106,651 *** -38,604 Δshareholder Wealth 0,0000891 *** 0,0001051 *** 0,0000476 *** 0,0000305 *** 0,0000345 *** 0,0000293 *** 0,0000165 *** 0,0000121 *** 0,0000366 *** Adjusted R-squared 0,086 0,042 0,052 0,024 0,065 0,065 0,032 0,025 0,063 F-value 109,905 *** 57,329 *** 76,448 *** 35,184 *** 98,508 *** 101,361 *** 50,306 *** 39,172 *** 102,235 *** Highest VIF 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 Cash sensitivity 0,000089 0,000105 0,000048 0,000031 0,000034 0,000029 0,000017 0,000012 0,000037 Sample 1152 1292 1383 1371 1407 1448 1481 1461 1499 2002 2003 2004 2005 2006 2007 2008 2009 2010 (1) (1) (1) (1) (1) (1) (1) (1) (1) Intercept 153,265 *** 88,372 *** 185,521 *** 79,793 *** -680,173 *** -83,296 ** -35,295 88,365 ** 324,569 *** Δshareholder Wealth 0,0000448 *** 0,0000565 *** 0,0000299 *** 0,0000435 *** -0,0000437 *** 0,0000622 *** 0,0000659 *** 0,0000166 * 0,0000736 *** Adjusted R-squared 0,085 0,058 0,014 0,027 0,006 0,064 0,064 0,001 0,042 F-value 145,556 *** 98,467 *** 23,710 *** 42,482 *** 9,018 *** 57,698 *** 102,360 *** 3,306 * 65,006 *** Highest VIF 1,000 1,000 1,000 1,000 1,000 1,000 1,100 1,000 1,000 Cash sensitivity 0,000045 0,000056 0,000030 0,000043 -0,000044 0,000062 0,000066 0,000002 0,000076 Sample 1560 1570 1561 1513 1427 1335 1483 1579 1448
27
Table 4b: Regression results hypothesis 2 – Model 1 Δ(salary + bonus) - step 2
1993 1994 1995 1996 1997 1998 1999 2000 2001 (2) (2) (2) (2) (2) (2) (2) (2) (2) Intercept 67,976 76,142 *** 26,643 *** 70,703 *** 67,937 *** 33,812 ** 93,770 *** 110,310 *** -38,316 Δshareholder Wealth t 0,0000942 *** 0,0001060 *** 0,0000473 *** 0,0000383 *** 0,0000390 *** 0,0000407 *** 0,0000201 *** 0,0000106 *** 0,0000368 *** Δshareholder Wealth t-1 0,0000164 * - 0,0000112 0,0000131 0,0000171 ** - 0,0000126 * - 0,0000281 *** - 0,0000168 *** - 0,0000038 - 0,0000011 Adjusted R-squared 0,088 0,042 0,052 0,027 0,067 0,077 0,045 0,026 0,063 R-squared Change 0,0026 0,0007 0,0008 0,0037 0,0024 0,0127 0,0130 0,0010 0,0001 F-value 56,723 *** 29,162 *** 38,827 *** 20,287 *** 51,171 *** 61,306 *** 35,582 *** 20,339 *** 51,171 *** Highest VIF 1,108 1,004 1,004 1,431 1,457 1,786 1,121 1,383 1,011 Non-cash sensitivity 0,000078 0,000117 0,000060 0,000021 0,000026 0,000013 0,000003 0,000007 0,000038 Sample 1152 1292 1383 1371 1407 1448 1481 1461 1499 2002 2003 2004 2005 2006 2007 2008 2009 2010 (2) (2) (2) (2) (2) (2) (2) (2) (2) Intercept 149,223 *** 89,151 187,678 84,913 *** -670,818 *** -62,104 -23,624 95,605 ** 293,046 *** Δshareholder Wealth t 0,0000471 *** 0,0000429 *** 0,0000302 *** 0,0000456 *** -0,0000451 *** 0,0000652 *** 0,0000731 *** 0,0000212 * 0,0000658 *** Δshareholder Wealth t-1 0,0000103 *** - 0,0000131 *** - 0,0000017 - 0,0000083 - 0,0000281 * - 0,0000246 ** - 0,0000280 *** 0,0000054 0,0000295 Adjusted R-squared 0,089 0,068 0,014 0,027 0,008 0,044 0,071 0,001 0,049 R-squared Change 0,0046 0,0098 0,0000 0,0011 0,0027 0,0039 0,0077 0,0002 0,0070 F-value 77,035 *** 57,930 *** 11,882 *** 22,091 *** 6,442 *** 31,657 *** 57,698 *** 1,845 38,041 *** Highest VIF 1,049 1,349 1,028 1,059 1,002 1,024 1,100 1,658 1,070 Non-cash sensitivity 0,000037 0,000030 0,000028 0,000037 -0,000073 0,000041 0,000101 0,000027 0,000095 Sample 1560 1570 1561 1513 1427 1335 1483 1579 1448
