The effect of the horizon problem on the CEO compensation structure

(1)

Amsterdam Business School

The effect of the horizon problem on the CEO compensation

structure

Name: Esmée Hilhorst Student number: 10835784

Thesis supervisor: prof. dr. V.R. O’Connell Date: 17 June 2016

Word count: 12,085

MSc Accountancy & Control, specialization Control

(2)

Statement of Originality

This document is written by student Esmée Hilhorst who declares to take full responsibility for the contents of this document.

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

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

(3)

Abstract

This research examines the effect of the horizon problem on the CEO compensation structure. Previous research found evidence that CEOs make different decision when their career horizon is short. Drawing on the agency theory and managerial opportunism, the compensation structure is a tool to align the interest of the CEO and the firm. Firms could control for the horizon problem in adapting the compensation structure of the CEO. Providing the CEO with more long-term incentives leads to an extension of the horizon of the CEO, because then a great amount of his rewards depend on it. In this paper, a sample is used consisting of 11,636 observations of US listed firms. Regression analyses are conducted to test the relation between the percentage of long-term incentives and the indicators for the horizon problem. The results suggest that when there are indicators of the horizon problem, CEOs receive less long-term incentives. This evidence is contrary to expectations and indicates that firms do not control for the horizon problem.

(4)

1 Introduction

1.1 Background

In this study the effect of the horizon problem on the compensation structure of CEOs is examined. CEOs usually are rewarded with a combination of cash-based compensation and long-term incentives. Cash-based compensations are basically salary and bonuses. Bonuses provide short-term incentives, because they are paid annually. Long-term incentives are mainly stock-based compensations. Most of the time, firms are interested in increasing the firm value and they want to provide the CEO with the incentive to increase firm value. They do this in providing CEOs with long-term incentives. Nevertheless, for managers the short-term results are just as important, because these results will determine the amount of bonus they receive at the end of the year. In most firms a so-called compensation or remuneration committee determines the CEO compensation structure. This committee needs to consist of independent members to make sure the decisions of the committee are objective and not influenced by the CEO.

For CEOs that are close to retirement or planning to leave the firm, the long-term results become less important, because they will stop working for the firm soon. When these events occur CEOs could go act in their own interest, which would be to try to establish an increase in short-term results instead of an increase in firm value in the future. A consequence can be that the CEOs will invest less money in new projects during their final years in the firm, to achieve better accounting numbers. If they succeed in this, they would get a higher bonus before they leave the firm. Nevertheless, investing in projects could have led to better results on the long term. CEOs that will leave in the near future have shorter career horizons. According to Smith and Watts (1982) this is called the horizon problem.

A way to force CEOs to focus on the future is the provision of long-term incentives. Yermack (1995) tried to find a relation between offering stock options awards and the CEO approaching retirement. The author expected to find that the firms increased stock options awards to incentivize the CEO when the CEO approached the age of 65. No evidence was found. The author researches compensation data from 1984 till 1991. These are the years just after the introduction of the horizon problem by Smith and Watts (1982), so it could be that companies were not even aware of the problem of short horizons at that time. Also, they do not take into account that CEOs do not retire at 65 often. When looking at the sample that is used for this research, it can be seen that a lot of CEOs are older than 65 and they are still working. This supports the assumption that a lot of CEOs do not retire at the age of 65.

(6)

firm. Therefore, when the interests of a CEO change, the firm should correct for this. Firms could control for the horizon problem by adapting the structure of CEO compensation when the CEO is close to the age of retirement or has plans to leave the firm. This study investigates if firms actually do this.

1.2 Research question

The research looks at the effect of short horizons in determining the compensation structure for CEOs. In particular, the study examines if firms take action to try to reduce the horizon problem by aligning the interests of the CEO and the shareholders. The research question would be: Do

firms change the CEOs compensation structure when the horizon problem is present?

1.3 Motivation

This research contributes to the existing literature in multiple ways. There is a gap in the literature in what firms do to control for the horizon problem or if they actually do something to reduce it. Several researches looked at the effect of the horizon problem, but they did not look at possible actions of the companies to prevent or solve it. Also, prior research examined the considerations when determining the compensation. Mostly, firm characteristics are taken into account, but there is no research available about the inclusion of CEO characteristics when designing the compensation.

1.4 Structure

In the study, an ordinary least squares analysis is conducted. First, the relevant literature is reviewed and discussed. Thereafter, the hypotheses are developed. Subsequently, the sample process is described and the methodology is explained. After that, the sample is analyzed and described. Section 5 presents and analyzes the results. Finally, the last section comprises the summary, conclusion and limitations of the study.

(7)

2 Literature

In this section, the relevant literature is reviewed. First, the agency theory and managerial

opportunism are explained. This theory is important to understand the use of compensation and incentives in motivating and guiding the CEO. Hereafter, several papers are discussed that research the effect of the compensation structure on the actions of CEOs. Also, the expectancy theory and the reinforcement theory are introduced. After that, literature about the horizon problem is reviewed. The phenomenon is explained and prior research about the effect of the horizon problem on several elements is disused. Furthermore, the key literature is summarized in table 10 in appendix 1.

2.1 The agency theory and managerial opportunism

Jensen and Meckling (1976) describe an agency relationship where there is a contract between the principal (owner of the firm) and the agent (controller of the firm). The principal delegates decision-making authority to the agent and the agent should act on behalf of the principal. Nevertheless, the agent will not always act in the best interest of the principal and thus act in his own interest. It is almost impossible to ensure that the agent will always make the most optimal decisions for the principal. This is called the agency problem. To prevent agency problems, interests have to be aligned. The principal should provide the agent with incentives to act in the principal’s best interest and make the best decisions for the welfare of the firm. One component of the solution to the agency problem is financial alignment, which indicates that the

compensation committee should provide the CEO with equity ownership and they should adapt the compensation structure when necessary (Jensen and Meckling, 1976; Fama and Jensen, 1983b).

Several researches referred to agency theory to explain the structure of CEO

compensation (Eisenhardt, 1985, 1989; Conlon and Parks, 1990). Compensation is used as a tool to align the interest of the firm with those of the CEO. Coles et al. (2006) examined the relation between managerial incentives and risk-taking. They found relationships between the executive compensation and the investment policy, debt policy and the firm risk. Coles et al. (2006) found evidence that when the wealth of the CEO is more dependent on stock volatility they show more risky behavior. This risky behavior expresses itself in more investments in R&D and fewer investments in PPE. In addition, they found that when the CEO adopts a more risky policy, most of the time this leads to a compensation structure where there are more long-term

(8)

between risk and the compensation structure of the CEO. They suggest that when firms operate in a more risky environment, the CEO is provided with more stock and option based

compensation. Moreover, they argue that when the risk of the firm changes firms will quickly adapt the compensation structure of the CEO. However, they claim that there will be a delayed impact of these changes in the portfolio of the CEO. Therefore, they conclude that the board should response quickly to future changes in the risk of the firm and they should react more excessively to these changes.

The agency theory indicates that CEOs behave opportunistically and therefore their interests have to be aligned with the interests of the shareholders. A theory that involves CEOs acting self-interested is managerial optimism. The theory about managerial optimism suggests that CEOs always act in their own interest. CEOs do not take the interest of the firm into account when they make decisions, but they always decide in their own benefit. As a result, it becomes a problem for the firm when the interests of the CEO and the shareholders differ from each other.

Devos et al. (2015) examined if CEOs behave opportunistic. They researched if CEOs exercise options before a split to take advantage of the stock price appreciation that occurs with a stock split. They found evidence that only a third of the CEOs of the sample exercised their options before the split announcement versus two third that exercised them after the split. This evidence suggests that CEOs take advantage of the stock price appreciation due to the split. Also Huddart and Lang (1996) looked at the exercise behavior of CEOs. They investigated the option exercise behavior of employees provided with long-term incentives. In their research they found evidence that the moment that CEOs exercise their options is consistent with financial factors as stock price and return, which indicates that they exercise their options when stock price and return are high. Both Devos et al. (2015) and Huddart and Lang (1996) confirmed the existence of optimistic behavior of CEOs.

Otto (2014) examined how optimism affects CEOs compensation. The proxies for optimism are based on the decisions about option exercises and the CEOs’ forecasts of the earnings per share. The author found that if there are indications of optimistic beliefs, the CEOs receive smaller stock options and fewer bonuses. They receive less total compensation as their peers. This evidence suggests that firms control for opportunism in determining the

compensation. However, he found that firms decrease the total compensation when such events occur and therefore he did not look at the explicit structure of the compensation. Also, the indications of optimistic beliefs are not based on personal characteristics of the CEO, but more on the previous actions of the CEO.

(9)

2.2 The effect of the CEO compensation structure

Two theories that look at designing optimal compensation structures to incentivize employees are the expectancy theory and the reinforcement theory. Vroom (1964) first came up with the

expectancy theory model. The theory suggests that providing employees with financial rewards based on their performance leads to more motivation and subsequent higher performance. Cammann and Lawler (1973) describe the model in their paper and discuss the three necessary components for a successful and effective incentive scheme. They claim that good performance should lead to rewards. These rewards have to be clear and achievable for the employees. Also, more positive outcomes need to come from the desirable performance than negative outcomes. This means that positive outcomes on one component should not lead to negative outcomes on another component. Cammann and Lawler (1973) conducted an experiment where they looked at the reactions of employees to an incentive plan. They examined if performance increased as predicted by the model when using it to come to the compensation structure. They found evidence that it did. The expectancy theory suggests that employees get motivated based on the promised rewards.

The reinforcement theory by Komaki et al. (1996) suggests that firms have to make very clear what behavior is wanted and what behavior is unwanted and act on this. The authors conducted an experiment where they rewarded employees with desirable behavior and punished the employees that showed undesirable behavior. They found that the average performance of the employees increased. This evidence suggests that reinforcing the consequences of behavior leads to an overall better performance.

Due to several changes, the combination of compensation for CEOs has changed compared to earlier. Clementi and Cooley (2009) investigated the compensation of CEOs in the US from 1993 to 2006. They found that, on average, the use of equity grants and the income of CEOs from the sale of stock have increased during that time.

As discussed in the section about the agency theory and managerial opportunism, CEOs act self-interested and they try to earn as much as they can. The compensation structure plays a big role in this. Drawing on the expectancy theory, it is important to design a fitting

compensation plan, because this leads to an increase in performance of the employees. Adapting the structure of the compensation can push the CEO in the right direction. Drawing on the reinforcement theory, when putting more weight on certain behavior and thus focus on explicit components of the compensation, the CEO will be incentivized to take certain actions.

According to Grossman en Hoskisson (2016), aligning the goals of the firm with those of the CEO in the design of the compensation also helps in gaining public confidence in the top

(10)

management of large public firms. They conducted research where they examined how the structure of the CEO compensation played a role in finding a balance in holding the CEO accountable and influencing the decision-making behavior of the CEO.

Several researchers found evidence that the actions of the CEOs depend on the

compensation structure. Efendi et al. (2007) examined incentives that led to the restatement of financial statements and found evidence that CEOs take advantage of opportunities to maximize their wealth. In their research they found that the chance of misstated financial statements increases when CEOs own a great amount of in-the-money stock options. Furthermore, they found that when the CEO is also the chair of the board, the chance of misstated financial statement also increases.

Gopalan et al. (2014) conducted research where they investigated the effects of short-term and long-short-term incentives on several components. They used pay duration as a deshort-terminant of the length of the incentives, because it can be difficult to establish if incentives are short-term or long-term. In their research they compared the pay duration to firm characteristics and to the presence of earnings management. To research the relationship between the duration of CEO pay and earnings management they used the level of discretionary accruals. As a result, they found a negative relation, which suggests that when incentives are more long-term the level of earnings management is lower. Furthermore, they found that CEOs who are rewarded on short-term results are more likely to use earnings-increasing accruals to increase the results.

Balsam and Miharjo (2007) examined the relation between equity and cash based compensation and voluntary CEO turnover. They found that providing the CEO with more equity based compensation decreases voluntary turnover. The same relationship is found with the amount of cash based compensation, except this effect is weaker. This evidence suggests that the chance of the CEO leaving the firm increases when there are more long-term incentives. The relationship between CEO turnover and short-term incentives is weaker and this would mean that long-term incentives have a greater effect on CEO turnover.

Dong et al. (2010) investigated the effect of providing CEOs with stock options on their risk taking behavior. According to Dong et al. (2010), providing the CEO with stock options helps in aligning the interests of the CEO and the shareholders. However, they state that the provision of stock options can even lead to too much risk taking by the CEO. This would be the case when the earnings of the CEO are too dependent on these stock options.

Berger et al. (1997) examined the effect of CEO entrenchment on the capital structure of the firm. They found that when the interests of the CEO and the shareholders are better aligned, the CEO accepts more debt. In addition, the CEO increases the leverage ratio to a more optimal

(11)

level. In conclusion, they argue that CEOs have to be forced to optimize the leverage level and this can be done by adapting the CEO compensation structure. These researches all support the assertion that CEOs act in line with the incentive design.

2.3 The horizon problem

There are various considerations when determining the compensation and researchers examined the effect of several different components. Carter et al. (2007) investigated the accounting role in determining the CEO compensation structure and looked at financial reporting concerns. As a proxy for the concerns they used the costs for financial reporting. They found a positive relation between financial reporting concerns and the use of stock options in the compensation structure and a negative relation between financial reporting concerns and the use of restricted stock in the compensation.

Boyd (1994) examined the effect of board control on the CEO compensation structure. He expected that CEOs would demand a higher salary when they had control over the board. However, he found that salaries were lower when there were higher control levels. This evidence is contrary to his expectations.

Daily et al. (1998) investigated the relationship between members of the compensation committee and the compensation structure. In particular they looked at independence of the committee members, where board members that could be influenced by the CEO were seen as not independent. They found no relationship between the independence of the members of the compensation committee and CEO compensation.

Furthermore, Belliveau et al. (1996) examined the effect of social capital on the CEO compensation structure. In the study, social capital was measured as status. They found a

negative relationship between the status of the chairman and the total CEO compensation, which means that when the chairman has a low status, the CEO receives higher compensation. In addition, a positive relation was found between the social capital of the CEO in comparison with the chairman and the total CEO compensation. No evidence was found for a relation between social similarity and the compensation of the CEO.

David et al (1998) investigated the influence of institutional investors on the structure of CEO compensation. They distinguish two types of investors, namely those with only an

investment relationship and those that depend on the firm for their own business. They found that only the institutional investors with an investment relationship could influence the

compensation structure of the CEO. They found evidence that those inventors were able to negatively influence the level of compensation and they were able to positively influence the

(12)

relative amount of long-term incentives in comparison with the total compensation.

One thing that can affect the agency relationship is the presence of the horizon problem. Smith and Watts (1982) first introduced the horizon problem. They state that the horizons of managers who are leaving the firm are short and explain what can happen (p. 145):

“The incentive effects of future salaries decrease as the manager is close to retirement. In the extreme, the sixty-four year old chief executive with one year’s service left will not be motivated by future salary adjustments.”

The horizon problem can arise in the case of retirement, but also when CEOs (have the intention to) leave to work for another firm. Normally, when CEOs leave the firm, the board already knows it long before they actually leave. CEOs that have the intention to leave will focus on the performance of their last year, because bonus compensation depends on that. Smith and Watts (1982) discuss how bonus plans for managers can affect the investment and financing decisions of the managers, especially for those with short horizons. Also, they suggest that the provision of long-term incentives can help in solving the horizon problem.

Prior research has covered the impact of CEOs short horizons on multiple variables. Kalyta (2009) looked at the impact of the horizon problem on earnings management. Moreover, he looked at pension plans that were based on performance and predicted that when there are such pension plans in place, it would be more likely that the CEO would indulge in earnings management. The proxy used for the horizon problem is managerial retirement. The author found a positive relationship between CEO retirement arrangements based on firm performance and income-increasing earnings management when CEOs were close to retirement.

Also Davidson et al. (2007) investigated the relationship between career horizons and earnings management. They argue that not only the retirement age leads to incentives to manage earnings, but also the structure of the compensation contributes to the horizon problem when a relatively large part of the total compensation is based on short-term results. In their research, they examine the effect of both CEO retirement and CEO compensation structure on earnings management. In particular, they focus on CEO turnover in combination with CEO age to

differentiate turnover and retirement. The presence of earnings management is measured through the amount of discretionary accruals. They found evidence that the amount of discretionary accruals is higher, when CEOs near retirement. In firms where CEOs retire and a larger part of the compensation is based on short-term results, they found a higher amount of discretionary accruals. However, this last result was not robust.

(13)

effect on future firm performance. For CEO tenure the evidence indicates that the results depend on the industry. In dynamic environments, higher CEO tenure decreases firm

performance, but in less dynamic environments this could actually increase firm performance. Moreover, McClelland et al. (2012) found results that CEOs with a short horizon perform worse than CEOs with a longer horizon. According to the authors, this worsened performance is caused by risk-averse behavior. They explain that when the CEO has a high level of ownership in the firm, this relationship becomes stronger. They argue that this is the case, since ownership comes with more power. For career horizon they used age as an indicator for short horizons. Furthermore, Gray and Cannella (1997) examined the relation between risk and the CEO compensation structure. They looked at the CEO compensation structure as a way to influence the behavior of CEOs and focused on the influence on risk taking behavior. In addition, they found evidence that the compensation structure is used as a tool to influence risk-taking behavior. Also, they stated that when a CEO nears retirement, this influences the type of investment decisions they make. This evidence supports that the horizon problem exists.

(14)

3 Hypotheses

In this section, the relationships that are expected are explained. First, the link between the horizon problem and the compensation structure is discussed. Thereafter, three hypotheses are formulated. The literature that is used to formulate the hypotheses is summarized in table 10 in appendix 1.

When CEOs have short horizons, they could act in their own interest. Agency problems will increase. To reduce the agency problems, CEOs have to be incentivized to act in the best interest of the stockholders and the firm. Therefore, a change in the compensation structure of the CEO should restore the agency relationship and this can prevent managerial opportunism. Firms can cut back the short-term incentives for CEOs with short horizons, so the short-term accounting numbers matter less for these CEOs. By providing the CEO with more long-term incentives, they stimulate the CEO in increasing the firm value. So, despite the fact that the CEO will leave the firm, it would still be remunerative for the CEO to invest in new projects.

Dechow and Sloan (1991) focused on the relationship between CEOs incentives and the investments that they make. They hypothesized that CEOs spend less money on discretionary investments in their final years, so they are able to increase the short-term results and maximize their end of year bonus. In their research they found that CEOs spend less on R&D and marketing during their final years in the firm. In addition, their results also indicate that CEO stock ownership mitigates the effect on R&D expenditures. When the CEO owns more stocks there is less reduction in expenditures. This evidence suggests that companies are able to control for the horizon problem by adapting the compensation structure.

Also Cheng (2004) found evidence for the positive effect of adaption of the CEO compensation structure on managerial behavior. He investigated the relationship between a change in R&D spending and changes in the CEO option compensation when the horizon problem is present. The proxy that is used for the horizon problem is when the CEO approaches retirement. The changes in compensation should prevent opportunistic behavior in the form of a decrease in R&D expenditures. He found that there is a positive relationship between changes in R&D expenditures and changes in CEO compensation when there are indications of the horizon problem. This evidence indicates that changing the CEO compensation structure is effective in reducing opportunistic behavior and upholding R&D spending and investments in projects.

Matta et al. (2008) investigated the implications of a shorter career horizon on risk taking in the form of an international acquisition. They found that CEOs approaching retirement with a

(15)

lot of equity holdings and in the money unexercised options are less likely to engage in

international acquisitions than CEOs that have less options and equity holdings. These results suggest that the business decisions of CEOs are influenced by the amount of options and equity that they own when they are close to retirement. In conclusion, they argue that despite owning options and equity, CEOs prefer having money instead of options. Therefore, they sell them as fast as they can. The researchers suggest that the companies, for example, should look at the expiration date of the options, rather than just giving more options to CEOs that are close to retirement. This evidence is contrary to Dechow and Sloan (1991) and Cheng (2004).

To examine the impact of short career horizons of CEOs on their compensation

structure, two hypotheses have been developed. I expect that when there are indications for short horizons, the compensation structure will be adapted to prevent managerial opportunism and reduce the horizon problem. When looking at the expectancy theory, the CEO compensation structure need to be adapted into a structure where the specific performance of CEOs is more linked to financial rewards (Vroom, 1964). To reduce the horizon problem, there need to be more weight on the link between an increase in firm value and the rewards on long-term incentives. Probably, this link is stronger and better known with CEOs when there are more long-term incentives provided for them. Drawing on the reinforcement theory, the reinforcement on long-term incentives should be bigger when dealing with the horizon problem (Komaki et al., 1996). This way, the CEO will have to focus on the long-term incentives, because a lot of his financial rewards depend on them. Providing more long-term incentives can extend their

horizon, but presumably this is only possible when a large amount of the financial rewards of the CEO depend on it.

For the first hypothesis the original way the horizon problem was explained by Smith and Watts (1982) is used. They had stated that the horizon problem arises when CEOs near retirement. So according to them, the problem arises with an older age. This leads to the following hypothesis:

H1: Firms use more long-term incentives in determining the compensation for CEOs when the CEO is older.

However, not only age can lead to a short horizon. When CEOs plan to leave the firm, this also could lead to the horizon problem. Therefore, the next hypothesis is as follows:

H2: Firms use more long-term incentives in determining the compensation for CEOs when the CEO has the intention to leave.

(16)

Another hypothesis is added to examine an interaction effect. It would be only logical that older CEOs are more likely to leave a firm than younger CEOs. Especially when they have already passed the retirement age. They have to retire one day. The following hypothesis is added to look at this interaction effect:

H3: CEOs that are older are more likely to have the intention to leave the firm.

Libby (1981) gives a framework that helps to illustrate a research. In the figure below, Libby’s framework is given adjusted to this research. The independent variable is measured through the proxies of CEO age (hypothesis 1) and CEO turnover (hypothesis 2). The dependent variable for the research is the compensation structure. This is measured through the relative amount of long-term incentives in relation to the total compensation. Six control variables are added to increase the validity of the outcomes. The control variables are: gender, firm size, return on assets (ROA), market-to-book ratio, z-score and firm industry. In the next section, all used variables are

explained in detail. Independent variable: Short horizons Dependent variable: CEO compensation structure Control variables: Gender, firm size, ROA, market-to-book ratio, z-score, firm industry Proxies: Relative amount of long-term incentives Proxies: - CEO age - CEO turnover

(17)

4 Sample and research methodology

In this chapter, the sample is described and the process of the selection of the sample is explained. Furthermore, the methodology is given, the formulas are presented and the used variables are discussed.

4.1 Sample selection

To research the effect of the horizon problem on long-term incentives, database research is conducted. Compensation data about listed firms in the US from 2004 until 2015 is gathered from Compustat ExecuComp. Financial data is gathered from Compustat IQ. The financial data is used to create control variables for the regression. These two datasets are merged into one.

Using the variable of the present age of the CEOs and the fiscal year (existing variables from ExecuComp), a total of 321 missing values in age are added. Hereby, only 10 observations had to be dropped because of missing values in age.

Eventually, only the data from 2004 to 2013 is used for the analysis. The years 2014 and 2015 are only used to create the dummy variable for CEO turnover. After the creation of the variable, the observations of 2014 and 2015 are dropped.

Furthermore, observations that contain missing values in variables that are used in the regression are dropped. Also variables that contain wrong information are dropped, such as a negative amount of long-term incentives. Each of the variables included is checked for missing values and specific errors. After correcting for this kind of errors, the remaining sample consists of 11,636 observations (table 1).

Table 1: Final sample after dropped observations

Number of observations

Begin sample 21,353

Less: Data from 2014 and 2015 2,015 Less: Missing values and negative

amounts in observations for creation of LTINC

452

Less: Missing values in observations for creation of MtoB

1,817 Less: Missing values in observations

for creation of ZSCORE

5,423 Less: Missing values in observation

for variable Age

(18)

4.2 Methodology

An ordinary least squares analysis is conducted to research the effect of a short horizon on the compensation structure. The dependent variable is the percentage of long-term incentives in comparison with the total compensation. The independent variables are two proxies for the horizon problem. Six control variables are added. The following formula is created:

𝐿𝑇𝐼𝑁𝐶 = 𝛽!+ 𝛽!𝑂𝐿10𝑃 + 𝛽!𝐶𝐸𝑂𝑇𝑂 + 𝛽!𝑂𝐿10𝑃 ∗ 𝐶𝐸𝑂𝑇𝑂 + β!𝑌𝑂10𝑃 + β!𝐺𝐸𝑁 + β_!𝑆𝐼𝑍𝐸 + β_!𝑅𝑂𝐴 + β_!𝑀𝑡𝑜𝐵 + β_!𝑍𝑆𝐶𝑂𝑅𝐸 + β_!!𝐼𝑁𝐷 + ε

LTINC = Percentage long-term incentives in relation to total compensation OL10P = 10% oldest CEOs (oldest 10% = 1, other = 0)

CEOTO = CEO turnover (leave next year = 1, leave in two years = 2) YO10P = 10% youngest CEOs (youngest 10% = 1, other = 0) GEN= Gender (male = 1, female = 0)

SIZE = Firm size (logarithm of total assets) ROA = Return on assets (net income/total assets)

MtoB = Market to book ratio (Market value/shareholder’s equity) ZSCORE = Z-score

IND = Firm industry group (1-digit SIC)

The two proxies for short horizons are: CEO turnover and CEO age. For both variables, dummies are created. CEO turnover comprises information about CEOs that left the next year or the year after. CEO age is split into two variables, one for the 10% oldest CEOs and one for the 10% youngest CEOs.

For the proxies CEOTO (CEO turnover) and OL10P (10% oldest CEOs) an interaction effect is expected. The oldest 10% of the CEOs will have already reached the retirement age, which increases the chance of them leaving the firm. Therefore, the interaction effect implies that when CEOs are older, the CEO turnover will be higher.

The variable SIZE is added to control for firm size. The variable consists of the logarithm of total assets. The larger the firm, the bigger the responsibilities will be for the CEOs. Therefore, the total compensation of the CEO of a large firm will be higher than for a CEO of a smaller firm. Wright et al. (2002) researched if compensation of CEOs changes when the firm size increases. They found that there is a positive relationship between firm size and CEO compensation. Also Dutta et al. (2011) found the same relationship.

Several researchers have shown that females are more risk averse than males (Khan and Vieito, 2013; Croson and Gneezy, 2009; Powell and Ansic, 1997). Their evidence suggests that companies should adapt the compensation structure to the gender of the CEO. When females are more risk averse, they should receive more long-term incentives to motivate them to invest in

(19)

new projects that can results in an increase in the firm value. Therefore, a control variable for gender is added (GEN).

The control variables ROA (return on assets) and MtoB (market to book ratio) are also added to the regression. Davila and Penalva (2006) claim that when CEOs are evaluated more on accounting performance measures, such as the return on assets and the market to book ratio, they generally receive more cash-based compensation. Therefore, in these firms there would be a positive relationship between the accounting performance measures and the received short-term incentives. However, Mehran (1995) found evidence for a positive relation between financial performance and long-term incentives. He used return on assets as a proxy for financial

performance and found a positive correlation between the return on assets and the percentage of compensation that is equity based.

The control variable ZSCORE is calculated according to the z-score formula of Altman1 (1968). The z-score can be used to calculate the credit risk of a company, which in other words stands for the possibility of bankruptcy. The lower the z-score, the higher the chance of

bankruptcy for a company. Gilson and Vetsuypens (1993) investigated the relationship between CEO compensation and bankruptcy. They found that companies adapt the compensation of CEOs when they are in financial trouble to incentivize them to reduce their financial problems. Also, they found that companies reduce the salaries and bonuses for CEOs in this kind of situations.

Furthermore, the industry of the firm has an impact on the CEO compensation. For example, investments are more important for investment companies than for other companies. The compensation of these companies would be more focused on investments in comparison with other companies. For this reason, the control variable IND is added. Kostiuk (1990) researched the relationship between firm size and firm industry and the compensation of

managers. He found that not only firm size has an impact on the compensation, but also industry characteristics have a significant impact on the incomes of managers. The variable IND is created based on the 1-digit SIC codes of the companies in the sample.

1_{Altman’s Z-score = 0.012X1 + 0.014X2 + 0.033X3 + 0.006X4 + 0.999X5}

X1 = working capital/total assets X2 = retained earnings/total assets

X3 = earnings before interest and taxes/total assets X4 = market value equity / book value of total debt

(20)

All the expected relations are summarized in the table below. For both the independent variables and the control variables, the predicted effects on the relative amount of long-term incentives in comparison to the total compensation are given.

Table 2: Predicted effects of the variables

In addition to the base regression, sensitivity analyses are conducted to test if the outcomes change when the independent variables are different. For the sensitivity tests, only the proxies for the horizon problem are changed. Instead of the oldest and youngest 10%, the oldest and

youngest 20% is used in the second regression. The following formula is created:

𝐿𝑇𝐼𝑁𝐶 = 𝛽_!+ 𝛽_!𝑂𝐿20𝑃 + 𝛽_!𝐶𝐸𝑂𝑇𝑂 + 𝛽_!𝑂𝐿20𝑃 ∗ 𝐶𝐸𝑂𝑇𝑂 + β_!𝑌𝑂20𝑃 + β_!𝐺𝐸𝑁 + β_!𝑆𝐼𝑍𝐸 + β_!𝑅𝑂𝐴 + β_!𝑀𝑡𝑜𝐵 + β_!𝑍𝑆𝐶𝑂𝑅𝐸 + β_!"𝐼𝑁𝐷 + ε

OL20P = 20% oldest CEOs (oldest 20% = 1, other = 0) YO20P = 20% youngest CEOs (youngest 20% = 1, other = 0)

Another sensitivity test is conducted, where the variable for CEO turnover is adapted. Instead of two dummies for the year before the CEO left (dummy = 1) and the year before that (dummy = 2), one dummy is created where 1 stands for the year before the CEO leaves and also the year before that. This variable is called CEOTO2. The following formula is used for the test:

𝐿𝑇𝐼𝑁𝐶 = 𝛽!+ 𝛽!𝑂𝐿10𝑃 + 𝛽!𝐶𝐸𝑂𝑇𝑂2 + 𝛽!𝑂𝐿10𝑃 ∗ 𝐶𝐸𝑂𝑇𝑂2 + β!𝑌𝑂20𝑃 + β!𝐺𝐸𝑁 + β_!𝑆𝐼𝑍𝐸 + β_!𝑅𝑂𝐴 + β_!𝑀𝑡𝑜𝐵 + β_!𝑍𝑆𝐶𝑂𝑅𝐸 + β_!"𝐼𝑁𝐷 + ε

Variable Predicted sign

CEOTO + OL10P + CEOTO*OL10P + YO10P - SIZE + GEN - ROA ? MtoB ? ZSCORE - IND ?

(21)

5 Descriptive statistics and empirical results

In this section, first a description of the statistics is given. Thereafter, the main regression model is checked for multicollinearity. One regression model is created to test whether the hypotheses are supported. Two extra regressions are created to test the sensitivity of the results. The results of the regression models are also discussed in this chapter.

5.1 Descriptive statistics

Table 3: Descriptive statistics for compensation data (n = 11,636)

Variables Mean Std. Dev. Min Max

Total compensation ($) 5923.804 7074.071 0.745 134457.9 Long-term incentives ($) 4669.454 6338.179 1.3315 129126.4 Short-term incentives (salary and bonus) ($)

1207.225 1781.678 0.001 77926

Percentage of long-term incentives (LTINC)

67.13% 24.4100 3.06% 100%

The total compensation of CEOs can be divided between short-term incentives, long-term incentives and other compensation. The total amount of long-term incentives is calculated through deducting salary, bonuses and other compensation from the total compensation. Other compensation comprises components that are not really short-term or long-term incentives, such as perquisites, tax related payments, severance payment and signing bonuses. On average CEOs receive $5.923.804 per year, where 67.13% of this amount consists of long-term incentives. The average amount of long-term incentives is $4.669.454.

(22)

Table 4: Descriptive statistics for CEO horizons (n = 11,636)

Age 55.7066 7.0808 28 96 YO10P (Youngest 10% = 1, other = 0) 0.1149 0.3189 0 1 OL10P (Oldest 10% = 1, other = 0) 0.0922 0.2893 0 1 YO20P (Youngest 20% = 1, other = 0) 0.2290 0.4202 0 1 OL20P (Oldest 20% = 1, other = 0) 0.1909 0.3920 0 1 CEOTO

(Leave next year = 1, leave in two years = 2)

0.2888 0.6226 0 2

The average age of the CEOs in the sample is the age of 55.71. As mentioned earlier, the 10% and 20% youngest and oldest CEOs are used for the regression. The mean for CEO turnover is 0.29.

Table 5: Descriptive statistics for control variables (n = 11,636)

SIZE 7.6599 1.5538 1.5518 12.7565 GEN (Male = 1, female = 0) 0.9705 .01691 0 1 ROA 0.0363 0.2982 -11.8493 24.0920 MtoB 3.2289 24.6813 -854.0903 1406.722 ZSCORE 2.7130 36.7158 -34.8756 2515.769 IND 4.0908 1.7175 1 9

The average logarithm of total assets (SIZE) is 7.66. Furthermore, 97.1% of the sample is male. The means for return on assets, market to book ratio and z-score are 0.04, 3.23 and 2.71 respectively. The average for industry is 4.09. If the industry codes are compared to the average amount and the percentage of long-term incentives, it can be concluded that both the amount and the percentage of long-term incentives does not really differ between industries. In the finance, insurance and real estate industry, the percentage of long-term incentives is a bit higher than the average. In the agriculture, forestry and fishing, and the non-classifiable companies, the percentage is a bit lower than the mean.

(23)

Table 6: Average amount and percentage of long-term incentives per industry (n = 11,636)

1-digit SIC code Industry Frequencies Average amount of long-term incentives ($) Average percentage of long-term incentives 1 Agriculture, forestry and fishing 42 4441.1966 59.99% 2 Mining and construction 827 5931.7537 69.45% 3 Manufacturing 5907 4558.0073 67.81% 4 Transportation, communications, electric, gas and sanitary service

974 5419.1376 63.86%

5 Wholesale trade and

retail trade

1585 4257.2645 65.45%

6 Finance, insurance and

real estate 339 4716.7868 70.44% 7 Services 1397 4614.949 67.27% 8 Public administration 535 3966.1945 65.58% 9 Non-classifiable 30 4117.8064 57.01% Average 4669.454 67,13% 5.2 Multicollinearity

To make sure the regression model is accurate, I test for multicollinearity by creating a correlation matrix. In appendix 2 (table 12), a correlation matrix is presented including the dependent, independent and control variables. As seen in the matrix, there is no sign of multicollinearity.

(24)

5.3 Regression analysis Table 7: Regression analysis (n = 11,636)

Variable Predicted sign Coefficient Standard error T-value P-value CEOTO Dummy = 1 + -2.9560 0.7431 -3.98 0.000*** Dummy = 2 + -2.1887 0.7803 -2.80 0.005*** OL10P + -7.8653 0.8865 -8.87 0.000*** CEOTO*OL10P 1*1 + -1.2595 1.8683 -0.67 0.500 2*1 + 1.3765 2.1627 0.64 0.524 YO10P - 0.8863 0.6612 1.34 0.180 SIZE + 6.0275 0.1357 44.40 0.000*** GEN - -1.5761 1.2280 -1.28 0.199 ROA ? 1.7500 0.6983 2.51 0.012** MtoB ? 0.0173 0.0084 2.06 0.039** ZSCORE - 0.0009 0.0056 0.16 0.874 IND ? 0.3052 0.1220 2.50 0.012** Intercept 22.2636 1.6975 12.87 0.000*** ***,**,*, Significant on a 1%, 5%, 10% level R-squared = 0.1626 Adj. R-squared = 0.1617

Table 6 presents the results of the regression. The adjusted r-squared for the regression is 16.17%, which means that 16,17% of the total variation is explained in the regression model.

With respect to CEO turnover, the results are contrary to my expectations. There is a negative significant relation between CEO turnover and the percentage of long-term incentives (t = -3.98; sig. = 0.000; and t = -2.8; sig. = 0.005). This means that in the years before the CEO leaves, the percentage of long-term incentives decreases. This effect is stronger in the year before the CEO actually left. Therefore, hypothesis 2 has to be rejected. However with caution, because it is not known for sure if the companies are aware of the intentions of the CEOs. An opposite effect is also found for the variable for oldest 10% CEOs. A negative relation is found between the oldest 10% of CEOs and the percentage of long-term incentives (t = -8.87; sig. = 0.000). This means, hypothesis 1 is rejected. For the variable of the youngest 10%, no significant relation is found.

The variables CEO turnover and oldest 10% are used for the interaction effect. As seen in the model, no significant interaction effect between the variables is found. Hereby, hypothesis 3 has to be rejected as well.

(25)

The control variables SIZE (t = 44.4; sig. = 0.000), ROA (t = 2.51; sig. = 0.012) and MtoB (t = 2.06; sig. = 0.039) are positively correlated with the percentage of long-term

incentives. For the variable SIZE, this is consistent with prior research. For ROA and MtoB no prediction could be made, but both relationships seems to be positive and significant on a 5% level. The results suggest no significant relationships for the control variables for gender (GEN) and z-score (ZSCORE).

Furthermore, the results for the control variable of IND implicate a positive relationship between IND and the percentage of long-term incentives (t = 2.50; sig. = 0.012). From this I conclude that firms that are more service oriented (1-digit SIC numbers above 4) provide their CEOs with relatively more long-term incentives.

To make sure there are no problems with multicollinearity between variables, an extra test is conducted in addition to the correlation matrix. This is a Variance Inflation Factor (VIF) test. The highest VIF value is 1.54 and the mean VIF value is 1.16, which confirms that there is no multicollinearity problem (appendix 3, table 13).

(26)

5.4 Sensitivity analyses Table 8: Sensitivity test (1) (n = 11,636)

Variable Predicted sign Coefficient Standard error T-value P-value CEOTO Dummy = 1 + -2.0224 0.8318 -2.43 0.015** Dummy = 2 + -1.6552 0.8657 -1.91 0.056* OL20P + -6.1366 0.6584 -9.32 0.000*** CEOTO*OL20P 1*1 + -1.9546 1.4717 -0.33 0.184 2*1 + 0.4808 1.6127 0.30 0.766 YO20P - 0.9948 0.5173 1.92 0.054* SIZE + 6.1001 0.1357 44.96 0.000*** GEN - -1.2942 1.2268 -1.05 0.291 ROA ? 1.7529 0.6974 2.51 0.012** MtoB ? 0.0175 0.0084 2.08 0.037** ZSCORE - 0.0007 0.0056 0.12 0.902 IND ? 0.2678 0.1219 2.20 0.028** Intercept 21.8120 1.7391 12.54 0.000*** ***,**,*, Significant on a 1%, 5%, 10% level R-squared = 0.1649 Adj. R-squared = 0.1641

For the second regression model, the variables of 10% oldest and 10% youngest CEOs are changed in variables for the 20% oldest and 20% youngest CEOs. The adjusted r-squared for this regression model is 16.41%, which means that 16,41% of the total variation is explained by this second regression model. This is 0.24% more than the first model.

The regression checks how sensitive the previous regression is on the created independent variables of age. In general, the regression gives the same results. The level of significant of some variables varies in this model compared to the previous model. Only one extra variable is significant in this model, this is YO20P. The relation is positive and significant on a 10% level (t = 1.92; sig. = 0.054). This would suggest that when the age is divided in bigger groups and the youngest 20% of the CEOs is compared to the compensation structure, a relation is would. The evidence implies that younger CEOs receive relatively more long-term incentives than older CEOS, which is contrary to the prediction.

The Variance Inflation Factor (VIF) test for this regression gives 1.85 as the highest VIF value and a mean VIF value of 1.27 (appendix 3, table 13). This confirms that there is no sign of

(27)

multicollinearity.

Table 9: Sensitivity test (2) (n = 11,636)

Variable Predicted sign Coefficient Standard error T-value P-value CEOTO2 + -2.5929 0.5632 -4.59 0.000*** OL10P + -7.8672 0.8865 -8.87 0.000*** CEOTO2*OL10P + -0.2657 1.5402 -0.17 0.863 YO10P - 0.8809 0.6612 1.33 0.183 SIZE + 6.0246 0.1357 44.38 0.000*** GEN - -1.5713 1.2280 -1.28 0.201 ROA ? 1.7642 0.6982 2.53 0.012** MtoB ? 0.0172 0.0084 2.05 0.040** ZSCORE - 0.0009 0.0056 0.16 0.875 IND ? 0.3036 0.1220 2.49 0.013** Intercept 22.2884 1.7301 12.88 0.000*** ***,**,*, Significant on a 1%, 5%, 10% level R-squared = 0.1624 Adj. R-squared = 0.1616

Table 8 presents the results of the regression where the variable for CEO turnover is changed into one dummy instead of two. The adjusted r-squared for the model is 16,16%. Compared to previous models, this model has less explanatory power. The model does not give different results or extra significant relationships.

Again there is no sign of multicollinearity. The highest VIF value is 1.72 and the mean VIF value is 1.15 (appendix 3, table 13).

(28)

5.5 Summary of the empirical results

As a summary of all the conducted regressions, the following table is given. All the created models are summarized and the findings are briefly discussed.

Table 10: Summary of results

Summary of regression

Hypothesis Summary Findings

Firms use more long-term incentives in determining the compensation for CEOs when the CEO is older

Two variables are created, namely 10% oldest and 10% youngest CEOs. The relation between these variables and the percentage of long-term incentives is examined

Negative significant relation between the oldest 10% of the CEOs and the percentage of long-term incentives

Firms use more long-term incentives in determining the compensation for CEOs when the CEO has the intention to leave

One variable is created for CEO turnover. The variable consists of two dummies. One dummy for the year before the CEO left (dummy = 1) and another for the year before that (dummy = 2). The relation between the variable and the percentage of long-term incentives is examined

Negative significant relation between CEO turnover and the percentage of long-term incentives. The effect is stronger in the year before the CEO leaves than two years before leaving CEOs that are older are more likely to

have the intention to leave the firm

The interaction effect between the variables of CEO turnover and the 10% oldest CEOs is investigated

No interaction effect between the variables is found

Summary of sensitivity analyses

Sensitivity analyses Summary Findings

Sensitivity test (1) The variables of 10% oldest and 10% youngest CEOs are changed into variables for the 20% oldest and 20% youngest CEOs

Overall, same results are found. One extra relation is significant in this model, this is the positive relation between the youngest 20% and the percentage of long-term incentives

Sensitivity test (2) The variable for CEO turnover is changed into one dummy for both one and two year before the CEO leaves instead of two separate dummies

Overall, same results are found. The model shows no extra significant relations

(29)

6 Summary and conclusion

In this last chapter, a short summary is given of the total research that is conducted. A conclusion is added where the contribution, the results and possible explanations for these results are

discussed. In the last part, the limitations of the study are provided.

6.1 Summary

In this research the effect of the horizon problem on the structure of the CEO compensation is examined. The horizon problem arises when the career horizon of a CEO shortens. Reasons for a shorter career horizon are, for example, when a CEO nears retirement and when a CEO has the intention to leave the firm. When these events occur, CEOs will stop working for the firm soon and therefore their interest in the future value of the firm will weaken. In conclusion, the horizon problem leads to a change in the interests of the CEO. When a CEO is not provided with a lot of long-term incentives, he will try to increase the short-term results because his bonus depends on this. Several researchers investigated the impact of short career horizons of CEOs on their actions and found that CEOs make different decisions when the horizon problem is present (e.g. McClelland et al., 2012; Gray and Cannella, 1997; Dechow and Sloan, 1991).

Drawing on the agency theory (Jensen and Meckling, 1976), the interests of the CEO have to be aligned with those of the firm. Providing the CEO with more long-term incentives is a way to align these interests, since it shifts the focus on short-term results to a focus on future firm value. Consequently, financial rewards of the CEOs are more dependent on long-term results. Looking at the expectancy theory (Vroom, 1964), there need to be more weight on the link between an increase in firm value and the rewards from long-term incentives to reduce the horizon problem. Also, the reinforcement of long-term incentives should be bigger when dealing with the horizon problem. Therefore, the reinforcement theory applies (Komaki et al., 1996). Taking all these theories together, providing the CEO with long-term incentives can extend the horizon of the CEO. However, this is only possible when a large amount of their financial rewards depend on these incentives.

Because adaption of the CEO compensation structure is a way to solve the horizon problem, the following research question is developed: Do firms change the CEOs compensation

structure when the horizon problem is present?

Two hypotheses are created to research the question. Another hypothesis is added to examine an interaction effect. The first hypothesis looks at the provision of long-term incentives when a CEO nears retirement:

(30)

H1: Firms use more long-term incentives in determining the compensation for CEOs when the CEO is older.

The second hypothesis is created with another indicator for the horizon problem, namely when a CEO plans to leave the firm:

H2: Firms use more long-term incentives in determining the compensation for CEOs when the CEO has the intention to leave.

The last hypothesis researches the interaction effect between the two proxies used for the horizon problem. I predict that when CEOs are older, the chance of turnover is higher:

H3: CEOs that are older are more likely to have the intention to leave the firm.

The research question is answered using an ordinary least square analysis. The dependent variable used is the percentage of long-term incentives in comparison with the total

compensation. The independent variables are two proxies for the horizon problem, namely CEO age (hypothesis 1) and CEO turnover (hypothesis 2). Two variables are created to investigate the effect of CEO age, namely 10% oldest and 10% youngest CEOs. For the creation of these variables, the 10% oldest and the 10% youngest CEOs are taken from the sample and dummies are created were CEOs that belong to these groups get a one. CEO turnover comprises a variable where two dummies are created, one dummy for the year before the CEO leaves and one for two years before the CEO leaves. Furthermore, six control variables are added that might affect the compensation structure (gender, firm size, return on assets, market to book ratio, z-score, firm industry).

The regression model gives contrasting results to my expectations. All hypotheses are rejected. The results suggest that when the indicators for the horizon problem are present, which means that CEOs are older or have the intention to leave, CEOs receive less long-term

incentives. This relation is found for both indicators. Also no significant interaction effect was found, which suggests that there is no significant relation between CEO age and CEO turnover.

Additional regression analyses are conducted to test the sensitivity of the results. In these models, the independent variables are changed. The proxies used to test CEO age, 10% oldest and 10% youngest CEOs, are changed into 20% oldest and 20% youngest CEOs. One extra relation is significant in this model compared to the first regression. For CEO turnover, the variable is changed into one dummy for the year before the CEOs leaves and two years before the CEO leaves. In the original regression, two different dummies were created for this. In general, these models give the same results.

6.2 Conclusion

(31)

determinants of CEO compensation are already investigated. Nevertheless, the horizon problem as a determinant of the compensation structure has not yet been researched extensively.

Smith and Watts (1982) first introduced the horizon problem and stated that the cause of the horizon problem is the retirement of a CEO. However, retirement is not the only reason why CEOs would leave the company. Still most researchers (e.g. Kalyta, 2009; Davidson et al., 2007; McClelland et al. 2012) only looked at retirement when examining the horizon problem. In this study, two indicators for the horizon problem are examined. CEO turnover is added to the regression, because CEOs change jobs for several more reasons than their retirement. They could go work for a different firm. The variable for CEO turnover leads to an extra contribution to the existing literature where the current literature is mainly focused on retirement.

The overall results contribute to the existing literature in showing what firms do to control for the horizon problem. Moreover, evidence is found that firm actually do not control for it. Instead of solving the excessive focus on short-term results by providing the CEO with more long-term incentives, firms adapt the structure in a way that could increase the horizon problem. When there are indicators for the presence of the horizon problems, firms provide CEOs with less long-term incentives than normal. Evidence is found for both the indicators of the horizon problem. When CEOs are older, they receive more long-term incentives in

comparison to their total compensation. In their final years at the firm, CEOs also receive more long-term incentives in comparison to their total compensation. These final years are not related to age. Moreover, in the last sensitivity test an extra significant relation is found. This is a positive relation between the 20% youngest CEOs and the relative amount of long-term incentives. The evidence from this regression indicates that when CEOs are younger they are provided with more long-term incentives in comparison to their total compensation.

There are possible explanations for the contrary results. It may be the case that firms just do not take the determinants of the horizon problem under consideration when creating the compensation structure. This would mean that firms do not know about the horizon problem and consequently, they are not aware of the evidence from prior research on the effect of the actions of CEOs. However, this would not explain the fact that the youngest CEOs receive more long-term incentives, which is odd. One possibility is that firms may be trying to tie the younger CEOs to the firm. Nevertheless, this relation is only significant when 20% of the CEOs of the sample are taken into account when creating the variable instead of 10% of the CEOs. Also the findings of this research raise questions about the independence of the compensation committee. It indicates that CEOs get exactly what they want and that may be a little too coincidental.

(32)

6.3 Limitations

Future research should look into the actual independence of members of the compensation committee and the possible influence of CEOs on this committee and their members. These further investigations may explain the contrary results and the lack in information decreases the explanatory power of the results. In addition, the results would have been stronger or could have been different if I were able to control for independence of the compensation committee.

Another suggestion for future research is the investigation of possible actions to decrease the horizon problem. Adaption of the compensation structure is already proven to be effective in aligning the interests of the CEO and the shareholder, but evidence in this study suggests that firms do not take this into account. However, there could be other ways to deal with agency problems and the horizon problem, such as performing more monitoring activities to check the actions of the CEO. Future research should focus on examining if firms take other actions to prevent the horizon problem.

Also, the effectiveness of adaption of the CEO compensation structure to prevent the horizon problem could depend on the kind of stocks that are provided to the CEO. For example, firms can choose to restrict the stocks that they provide. As discussed in the literature review, Carter et al. (2007) found contrary results in their investigation of the relation between financial reporting concerns and the provision of stock options. They found a positive relation for stock options and they found a negative relation for restricted stock options. The data for my research was gathered from Compustat ExecuComp, but not all information is available in this database. The database makes no distinction between the kinds of stocks that the CEO is

provided with. If this distinction could be made, the compensation structure could be determined more precisely. This could lead to a more in depth research of the effects of CEO compensation structure on actions of the CEO. Also, the actions of the firms in controlling for the horizon problem could be examined in more depth.

Unfortunately, a lot of observation had to by dropped in this study due to incomplete variables in the sample. In the beginning the sample consisted of 21,353 observations and after several checks 11,636 observations remained. This means approximately 46% of the observations had to be dropped, which made the sample a lot smaller. A larger sample would have increased the reliability of the findings.

(33)

References

Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The journal of finance, 23(4), 589-609.

Balsam, S., & Miharjo, S. (2007). The effect of equity compensation on voluntary executive turnover. Journal of Accounting and Economics, 43(1), 95-119.

Belliveau, M. A., O'Reilly, C. A., & Wade, J. B. (1996). Social capital at the top: Effects of social similarity and status on CEO compensation. Academy of management Journal, 39(6), 1568-1593.

Berger, P. G., Ofek, E., & Yermack, D. L. (1997). Managerial entrenchment and capital structure decisions. The journal of finance, 52(4), 1411-1438.

Boyd, B. K. (1994). Board control and CEO compensation. Strategic Management Journal, 15(5), 335-344.

Cammann, C., & Lawler, E. E. (1973). Employee reactions to a pay incentive plan. Journal of Applied Psychology, 58(2), 163.

Carter, M. E., Lynch, L. J., & Tuna, I. (2007). The role of accounting in the design of CEO equity compensation. The Accounting Review, 82(2), 327-357.

Cheng, S. (2004). R&D Expenditures and CEO Compensation. The Accounting Review, 79(2), 305-328.

Clementi, G. L., & Cooley, T. F. (2009). Executive compensation: facts (No. w15426). National Bureau of Economic Research.

Coles, J. L., Daniel, N. D., & Naveen, L. (2006). Managerial incentives and risk-taking. Journal of

financial Economics, 79(2), 431-468.

Conlon, E. J., & Parks, J. M. (1990). Effects of monitoring and tradition on compensation arrangements: An experiment with principal-agent dyads. Academy of Management

journal, 33(3), 603-622.

Croson, R., & Gneezy, U. (2009). Gender differences in preferences. Journal of Economic

literature, 47(2), 448-474.

Grossman, W., & Hoskisson, R. E. (1998). CEO pay at the crossroads of Wall Street and Main: Toward the strategic design of executive compensation. The Academy of Management

Executive, 12(1), 43-57.

Daily, C.M., Johnson, J.L., Ellstrand, A.E., & Dalton, D.R. (1998). Compensation committee composition as a determinant of CEO compensation. Academy of Management Journal 41(2), 209-220.

(34)

David, P., Kochhar, R., & Levitas, E. (1998). The effect of institutional investors on the level and mix of CEO compensation. Academy of Management Journal, 41(2), 200-208.

Davidson III, W. N., Xie, B., Xu, W., & Ning, Y. (2007). The influence of executive age, career horizon and incentives on pre-turnover earnings management. Journal of Management &

Governance, 11(1), 45-60.

Davila, A., & Penalva, F. (2006). Governance structure and the weighting of performance measures in CEO compensation. Review of Accounting Studies, 11(4), 463-493.

Dechow, P. M., & Sloan, R. G. (1991). Executive incentives and the horizon problem. Journal of

Accounting and Economics, 14(1), 51-89.

Devos, E., Elliott, W. B. & Warr, R. S. (2015). CEO opportunism?: Option grants and stock trades around. Journal of Accounting and Economics, 60(1), 18-35.

Dong, Z., Wang, C., & Xie, F. (2010). Do executive stock options induce excessive risk taking?

Journal of Banking & Finance, 34(10), 2518-2529.

Dutta, S., MacAulay, K., & Saadi, S. (2011). CEO power, M&A decisions, and market reactions.

Journal of Multinational Financial Management, 21(5), 257-278.

Efendi, J., Srivastava, A., & Swanson, E. P. (2007). Why do corporate managers misstate financial statements? The role of option compensation and other factors. Journal of Financial

Economics, 85(3), 667-708.

Eisenhardt, K. M. (1985). Control: Organizational and economic approaches. Management

science, 31(2), 134-149.

Eisenhardt, K. M. (1989). Agency theory: An assessment and review. Academy of management

review, 14(1), 57-74.

Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. The Journal of Law &

Economics, 26(2), 301-325.

Gilson, S. C., & Vetsuypens, M. R. (1993). CEO compensation in financially distressed firms: An empirical analysis. The Journal of Finance, 48(2), 425-458.

Gopalan, R., Milbourn, T., Song, F., & Thakor, A. V. (2014). Duration of executive compensation. The Journal of Finance, 69(6), 2777-2817.

Gormley, T. A., Matsa, D. A., & Milbourn, T. (2013). CEO compensation and corporate risk: Evidence from a natural experiment. Journal of Accounting and Economics, 56(2), 79-101. Gray, S. R., & Cannella, A. A. (1997). The role and risk in executive compensation. Journal of

Management, 23(4), 517-540.

Huddart, S., & Lang, M. (1996). Employee stock option exercises an empirical analysis. Journal of

(35)

Jensen, M. C., & Meckling, W. H. (1976). Theory of the Firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.

Kalyta, P. (2009). Accounting discretion, horizon problem, and CEO retirement benefits. The

Accounting Review, 84(5), 1553-1573.

Komaki, J., Waddell, W. M., & Pearce, M. G. (1977). The applied behavior analysis approach and individual employees: Improving performance in two small businesses. Organizational

Behavior and Human Performance, 19(2), 337-352.

Khan, W. A., & Vieito, J. P. (2013). CEO gender and firm performance. Journal of Economics

and Business, 67, 55-66.

Kostiuk, P. F. (1990). Firm size and executive compensation. Journal of human Resources, 25(1), 90- 105.

Libby, R. (1981). Accounting and human information processing: Theory and applications. Michigan: Prentice Hall.

Matta, E., & Beamish, P. W. (2008). The accentuated CEO career horizon problem: Evidence from international acquisitions. Strategic Management Journal, 29(7), 683-700.

McClelland, P. L., Barker, V. L., & Oh, W. Y. (2012). CEO career horizon and tenure: Future performance implications under different contingencies. Journal of Business Research, 65(9), 1387-1393.

Mehran, H. (1995). Executive compensation structure, ownership, and firm performance. Journal

of financial economics, 38(2), 163-184.

Otto, C. A. (2014). CEO optimism and incentive compensation. Journal of Financial Economics,

114(2), 366-404.

Powell, M., & Ansic, D. (1997). Gender differences in risk behaviour in financial decision- making: An experimental analysis. Journal of economic psychology, 18(6), 605-628.

Smith, C. W., & Watts, R. L. (1982). Incentive and tax effects of executive compensation plans.

Australian Journal of Management, 7(2), 139-157.

Vroom, V. H. (1964). Work and Motivation. New York: Wiley & Sons.

Wright, P., Kroll, M. & Elenkov, D. (2002). Acquisition Returns, Increase in Firm Size, and Chief Executive Officer Compensation: The Moderating Role of Monitoring. The Academy of

Management Journal, 45(3), 599-608.

Yermack, D. (1995). Do corporations award CEO stock options effectively? Journal of Financial

The effect of the horizon problem on the CEO compensation structure