What is the relationship between the compensation package of a CEO and the firma performance within the US banking industry? : a comparison between the financial crisis and the period after

(1)

What is the relationship between the compensation

package of a CEO and the firm performance within the

US banking industry? A comparison between the

financial crisis and the period after.

Profijt, Nadine

10747796

31 January 2018

Supervisor: R. Gardner

Bachelor in Finance and Organisation

(2)

Statement of Originality

This document is written by Nadine Profijt 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 paper investigates the relationship between the compensation package of a CEO and the performance of a firm, within the US banking industry. In contrast to previous research, this study will also make a comparison, for this relationship, between the period of the financial crisis and the period after. The effects of five separate

components of a CEO compensation package will be determined, namely: salary, bonus, stock awards, option awards and equity-‐based awards. The Tobin’s q is used to measure the bank performance. This research performed a multiple regression analysis on the Tobin’s q and the five different components of a CEO compensation package, using panel data. The sample included 154 US commercial banks and was taken from the time

period of 2005 until 2016. The results did not show enough statistical evidence to prove that each of the five separate components of a CEO compensation package has an effect on the performance of a US bank. Furthermore, the results did also not show enough statistical evidence to prove that; the effect of these separate components on the firm performance was different during the financial crisis.

(4)

1. Introduction ... 5

2. Literature Review ... 7

2.1 Agency Theory ... 7

2.2 Measuring Executive Compensation ... 8

2.3 Measuring Firm Performance ... 10

2.4 Former research ... 11

2.4.1 Executive Compensation and Firm Performance ... 11

2.4.2 Executive Compensation and Firm Performance:

The Financial Crisis ... 13

2.5 Hypotheses ... 14

3. Data & Methodology ... 16

3.1 Data ... 16

3.2 Model and regression ... 17

3.3 Descriptive statistics ... 20

4. Results and Analyses ... 22

5. Conclusion and Discussion ... 26

6. References ... 28

(5)

1. Introduction

For more than a decade there has been an ongoing debate about the growth of executive compensation, which increased the interest in the subject of CEO compensation

packages. Especially executive compensation packages in relationship with the

performance of a firm, has raised a lot of attention. High executive compensation should cause a great firm performance in the eyes of many, however this is not always the case. For example, in 1990 Stephen M. Wolf, the CEO of UAL Corporations, which was the parent company of United Airlines, was paid a total compensation of Eighteen million three hundred one thousand dollars (Barris, 1992). In that same fiscal year, the

performance of United Airlines was from poor quality. It had only a profit of $95 million, which was a 70% drop in comparison with the previous year (Barris, 1992).

Because of the growth in executive compensation, the whole subject about executive compensation in relationship with firm performance drew a lot of attention among researchers. For instance, Mehran did research to this relationship in 1995. Mehran (1995) found a positive relationship between CEO compensation and firm performance. This examination provided evidence determining that the form of

compensation is more important than the level of compensation, for the motivation of a manager to increase the firm performance (Mehran, 1995). Thus, Mehran (1995) not only looked at the total compensation package, he did also pay attention to the different components of such a package. Mehran (1995) concluded that firm performance and executive compensation had a positive relationship. The majority of studies found the same conclusion about the relationship between a compensation package of a CEO and the performance of the firm (Hall and Liebman, 1998, and, Core, Holthausen and

Larcker, 1999, among others). However, the research of Bebchuck and Fried (2005) was one of the studies that did not found a positive relationship between executive

compensation and firm performance.

Also, the financial crisis in 2007 and 2008 drew a lot of attention to the executive compensation-‐firm performance relationship. Researchers tried to find the causes, which provoked the banking industry collapse, especially within the US. An outstanding argument found for the collapse, was that the CEO’s incentives were too weak.

According to Blinder (2009), this argument is one of the most crucial causes for the collapse within the US banking industry. The executives’ poor incentives were mainly

(6)

caused by the fact that the compensation schemes of the CEO’s were not accurately aligned with the performance of the banks (Fahlenbrach & Stulz, 2011).

Based on the prior research, there can be concluded that many researchers studied about the relationship between a CEO compensation package and the

performance of a firm. This relationship was either linked with the financial crisis or linked to another period in time. However, none of these researchers made a

comparison of the executive compensation-‐firm performance relationship between a normal period in time and the financial crisis period. That is the reason why this paper makes a comparison, for this relationship, between the financial crisis and the period after. Therefore, the research question is: What is the relationship between a

compensation package of a CEO and the firm performance within the US banking industry? A comparison between the financial crisis and the period after. Several components of a CEO compensation package will be used as independent variables and the firm

performance will be used as the dependent variable. In this way, the effect of one separate compensation package component on firm performance can be determined. Using different components solely, as independent variables, is a unique way to study this relationship. Most studies focused on the effect of total compensation on the firm performance or focused on total compensation in combination with different

components of the compensation package. However in this thesis, the precise effect of the separate components of a compensation package on the firm performance will be determined and the effect of the total compensation will be omitted. For instance, whenever the effect of the total compensation on firm performance turns out to be negative, one may be interested which component(s) of a compensation package causes this negative effect. Furthermore, the effect of these separate compensation package components can be compared between the period of the financial crisis and the period after. The US banking industry is used, because former research never delved deep into this industry when determining the executive compensation and firm performance relationship. However, this industry suffered hard under the financial crisis and it is thus interesting to search for differences between the effect of the separate

compensation package components on firm performance in the financial crisis and the period after.

This thesis is structured as follows. In the second section a literature review about the executive compensation and its underlying theory is given. Also the measurements of

(7)

CEO compensation and firm performance are described within the literature review. Furthermore, some former research on the relationship between firm performance and CEO compensation is required in this section. The last part of the second section gives the hypotheses made, based on the literature review. The third section provides an elaborate clarification of the data and methodology used. In the fourth section the

regression results will be given and an analyses is made, based on these results. The fifth and last section contains a conclusion made about the research, based on the other four sections.

2. Literature Review

2.1 Agency Theory

To understand executive compensation, first the underlying agency theory has to be explained. The agency theory is a subject that has been studied for a long time now. Agency theory is a theory based on the relationship between an agent and its principal. In 1976 Jensen and Meckling were one of the first’ who gave a precise definition of this relationship which is defined by a contract that one person, the principal, wants another person, the agent, to provide services under the principal’s supervision. Jensen and Mekcling also state that within the agency relationship, the agent will get some decision-‐ making authority appointed from the principal. According to Ross (1973), almost all examples of agency are universal, because all contractual agreements between employers and employees can generally be seen as an agency relationship.

Whenever both the principal and the agent want to maximize their utility, there is a chance that the agent will not act in the best interests of the principal all the time (Jensen and Meckling, 1976). This problem, where the principal-‐agent relationship is disturbed, can also be referred to as the agency problem. If an agency problem occurs, there may be some losses and there may also be some costs of monitoring the activities of the agent that are incurred by the principal (Tosi and Gomez-‐Meija, 1989). These costs and losses are often referred to as agency costs. According to Eisenhardt (1989), the agency theory is constructed to fix the problems that can arise within a principal-‐ agent relationship. There are two problems that this theory has the potential to resolve:

(8)

(a) the problem which arises when the interests and goals of the principal and the agent conflict and (b) the problem that it is expensive and hard for the principal to check whether the agent is actually doing what the principal authorized (Eisenhardt, 1989). The main idea of the agency theory is that setting a contract may be the most efficient way to cover the principal-‐agent relationship (Eisenhardt, 1989). Chen and Jermias (2014) suggest that because of the diverging interests of the principal and the agent, it may be possible to assume that the agent will act ‘selfish’ and will not act in the interest of the agent, but use their information advantage to maximize their own interests. Therefore, the agency theory should construct a contract at where there is an incentive plan that will link the compensation of the agent to the firm’s performance (Chen and Jermias, 2014). According to Eisenhardt (1989), this contract relies on human

assumptions, organizational assumptions and information assumptions. It is important to figure out how this contract can be made as efficient as possible.

For this thesis, the principal-‐agent relationship is between the CEO and the stockholders of American banks. In these situations, the CEO is the agent and the stockholders of the bank are the principal. According to the agency theory, in order to prevent agency problems, there should be made an efficient contract at which both the principal and the agent maximize their utility. In such an efficient contract the

compensation package of the CEO will be determined. The composition of the compensation package of the CEO is often referred to as executive compensation.

2.2 Measuring Executive Compensation

According to Murphy (2012) executive compensation can be seen as an assumption about how to measure the total compensation that a CEO will receive. But the changes and differences in executive compensation in cross-‐country, cross-‐sectional has been an ongoing debate. This total compensation of a CEO will not only rely on base salary that is set at the beginning of each year. Otherwise, it would be easy to compare salaries across executives, or to compare salaries across different years to see whether and why there have been changes over the years (Murphy, 2012). However, executives receive a compensation package which is a mix of a lot of aspects and which differs for every company. This mix can contain base salaries, performance shares, restricted stock, stock options, stock awards, cash bonuses, long-‐term incentives, retirement benefits and other

(9)

compensation (e.g. health benefits, club memberships etc.) (Murphy, 2012). In this thesis a few of these aspects are used to measure the executive compensation package. The first aspect used, will be the base salary, a component of the compensation package, which is seen as the most fixed part (Tosi et al, 1989). According to the research of Dittman and Maug (2007), the base salary should be low in the

compensation scheme of an executive. However, according to Mehran (1995) base salaries should be low, because of the risk aversion of CEO’s. There has been done research, and top managers are generally risk averse and thus prefer to have as less risk as possible (Mehran, 1995). This means that they would want their compensation schemes structured in a way that they bear less personal risk (Harris and Raviv, 1979). Mehran (1995) stated that executives should thus prefer higher fixed salaries, because of the less personal risk it contains.

The second component within the CEO compensation package, which is used in this thesis, is the bonus. According to Tosi et al (1989) bonuses are defined as cash awards to the executives in a short time period. However, Healy (1983) finds that the definitions used in bonus schemes always differ between companies. Companies can, for example, award a big bonus whenever a CEO provides a superior performance. Also, the firm can award a bonus per performance output. According to Healy (1983) companies can also use bonus plans based on accounting earnings to compensate its executives. However, according to Murphy and Jensen (1990) at most companies the bonus is a big part of the compensation structure and this will thus not generate a lot of fluctuations within the compensation scheme of the executive.

According to Baker, Jensen and Murphy (1988), economic models, which predict compensation schemes, assume that a higher performance will require a greater effort. These models state that a compensation scheme should have parts in it at which the expected utility of a worker increases with observed productivity (Baker, Jensen and Murphy, 1988). Thus, there are performance measurements within a compensation scheme at which the pay is directly based on the performance of a worker: pay-‐for-‐ performance system. The third and fourth components used in this thesis are such pay-‐ for-‐performance systems. With stock and option awards an executive’s pay is based on the equity of the firm (Mehran, 1995). According to Coughlan and Smith (1985), stock and option awards give the CEO an incentive to act in ways that will increase the value of the stockholders in that year, because the executives value of options and stock they

(10)

are holding will then increase as well. However, Narayanan (1996) proved that stock based compensation, as only compensation component, is not efficient enough. Jensen (2005) suggests that stock-‐ and option-‐based rewarding, especially option based, have a damaging effect on the performance of a firm, because it encourages the executives to pay too much attention to increasing the short-‐term stock price. On the other hand stated Dittman and Maug (2007) that stronger and more efficient contracts would include smaller amounts of options and many more stock awards. There has thus been done a lot of research, to this component of a compensation package, all with different outcomes.

2.3 Measuring Firm Performance

Several studies show that there is a wide array of measures to determine firm

performance with respect to executive compensation. In this thesis Tobin’s q is used to measure the performance of a firm. According to Chung and Pruitt (1994) Tobin’s q plays a role that is important in numerous financial interactions. Tobin’s q can be

defined as the ratio of the market value of a corporation to the value of its assets (Chung and Pruitt, 1994). This performance measure explains many diverse corporate

symptoms according to different researchers (Chung and Pruitt, 1994). Smith and Watss (1992) suggest that Tobin’s q explains the relationship between financial measures, dividend and compensation policies. It can also explain the relationship between the equity ownership of managers and the value of the firm according to McConnell and Servaes (1990). Chung and Pruitt (1994) took the assumptions of the former

researchers into account and came up with the following approximation:

Tobin’s q= (MVE+PS+DEBT)/TA. Where MVE is the firm its share price plus the number of shares outstanding of the common stock of a firm, PS measures the value of the outstanding preferred stock of a firm that can be liquidated (Chung and Pruitt, 1994). DEBT can be explained as the value of the firm’s short-‐term liabilities and the firm its book value of the long-‐term debt, and TA can be defined as the book value of the total assets of the corporation. According to Chung and Pruitt (1994) this approximation has on both academic and financial field a significant interest.

(11)

2.4 Former research

2.4.1 Executive Compensation and Firm Performance

There has been done a lot of research to the relationship between executive

compensation and firm performance. In 1995 Mehran was one of the first’ who did research to this relationship. His research served as an important fundament for other researchers to build on. Mehran (1995) examined the compensation structure of 153 manufacturing firms in 1979-‐1980, which were all randomly selected. This examination provided evidence, which supported incentive compensation and which also,

determined that the form of the compensation is more important, than the level of compensation, for the motivation of a manager to increase the firm performance (Mehran, 1995). Mehran’s (1995) paper provides empirical evidence that the

performance of a firm has a positive relationship with the percentage of the managers compensation that is equity-‐based. Examples of equity-‐based compensation are: stock options, stock awards, option awards and preferred stock. He used firm performance as the dependent variable and he measured it with Tobin’s q and the ROA (return on assets). The independent variables, three measures of compensation, he used were: new stock options as a percentage of the total compensation, the percentage of the total compensation that is based on equity and the salary plus bonus as a percentage of the total compensation.

Three years later, Hall and Liebman (1998) did a comparable empirical research to this relationship as well. However, they focused solely on a fifteen-‐year panel date of the largest, publicly traded firms from the U.S. (Hall and Liebman, 1998). Hall and Liebman (1998) found a strong correlation between CEO compensation and the

performance of a firm. According to Hall and Liebman (1998), the change in value of the executives holdings in stock options and general stock, is the base of the generated relationship between firm performance and CEO compensation. The correlation between firm performance and executive pay has risen since 1980, because of the increase in stock and stock option awards (Hall and Liebman, 1998).

Core, Holthausen and Larcker did in 1999 also an empirical research to the relationship between firm performance and the compensation package of a CEO, with

(12)

the aid of Mehran’s, Hall’s and Liebman’s former research. Nevertheless, they studied if the corporate governance of a firm (board and ownership structure) also correlated in this relationship. They found that a significant part of the cross-‐sectional variation in executive compensation is explained by measures of the corporate governance of a firm (Core, Holthausen and Larcker, 1999). Core, Holthausen and Larcker (1999) found that, whenever the structure of a firm its corporate governance is weak, the CEO

compensation is greater. This greater CEO compensation, raised by the less effective corporate governance structure, has a statistically negative relationship with the firm performance, according to the research of Core, Holthausen and Larcker (1999). With their results, Core, Holthausen and Larcker (1999) concluded that a firm has bigger agency problems when its corporate governance structure is less effective; that executive compensation is higher at corporations with greater agency problems and; that the performance of a firm is worse when there are greater agency problems at that firm.

In 2008 Canarella and Gasparyan also examined the relation between firm performance and executive compensation. However, they also studied whether the firm size is related to firm performance and CEO compensation. Canarella and Gasparyan (2008) took a panel of firms from the USA in the period between 1996 and 2002. They concentrated on total CEO compensation, which included equity-‐based awards, and they took two measures for firm performance: total shareholders return and return on assets (Canarella and Gasparyan, 2008). Canarella and Gasparyan (2008) found empirical evidence that the CEO compensation is positively and significantly related to the size and the performance of a firm. However, they found some evidence that the effect of firms size on CEO compensation is more significant in the second period than in the first one (Canarella and Gasparyan, 2008). Finally, Canarella and Gasparyan (2008) found that in the second period, when the stock market crashed, both firm performance measures affected the total CEO compensation. The stock market crash was the beginning of the financial crisis, which also had an effect on the relationship between CEO compensation and firm performance. This will be discussed in the following part.

(13)

2.4.2 Executive Compensation and Firm Performance: the Financial Crisis

Since the financial crisis of 2007 there has been done further research to the influence of CEO compensation on firm performance. Most researchers conclude that there is a strong link between de executive pay and the bad performance of the firm, within the financial crisis.

In 2011 Fahlenbrach and Stulz investigated if the incentives of executives, before the credit crisis, are related to bank performance during the crisis. The paper illustrates evidence that CEO’s, who’s incentives were better aligned with the interests of the firm, performed worse and evidence that CEO’s performed better were not found

(Fahlenbrach & Stulz, 2011). Fahlenbrach and Stulz (2011) also found evidence for the fact that option compensation and cash bonuses payment did not cause worse

performance for banks during the credit crisis. CEO’s took risks before the crisis and, according to Fahlenbrach & Stulz (2011), whenever an executive takes a risk that is not aligned with the interest of the shareholder, this executive should sell some shares. However, before and during the crisis, executives of banks did not decrease their share holdings, which caused extremely high losses of wealth during the crisis (Fahlenbrach & Stulz, 2011).

Mehran, Morrison and Shapiro analyzed in 2011 the problems in the governance structure exposed by the financial crisis. Since 2006, executives in the banking industry have had the most compensation of all CEO’s in the entire economy (Mehran, Morrison & Shapiro, 2011). According to Mehran, Morrison and Shapiro (2011), financial

institutions where executives had more incentives to take risk, performed worse during the credit crisis. Thus, the higher the level of a CEO’s risk-‐taking, the higher the volatility of a firm (Mehran, Morrison and Shapiro, 2011). According to Mehran, Morrison and Shapiro (2011), the CEO compensation should be partly tied to a measure of the default riskiness of a corporation. In this way, the objective of an executive will be more aligned with the social objectives and interests related to risk choice.

In 2014, Bhagat and Bolton also did research to executive compensation during the financial crisis. This paper studied the structure of executive compensation, in the 14th largest financial institutions within the U.S., during 2000 until 2008 (Bhagat & Bolton, 2014). Bhagat and Bolton (2014) focus on the purchases and sales of a CEO’s bank stock, the CEO’s salary and bonus, and the capital losses the executives made

(14)

because of the impressive reduction of the share price in 2008. According to Bhagat and Bolton (2014), the excessive risk-‐taking of financial institutions is sufficiently correlated with the incentives provoked by the CEO compensation structure. The results and conclusions of Bhagat and Bolton (2014) are not aligned with those of Fahlenbrach and Stulz (2011), because Bhagat and Bolton do not support that the poor bank performance was generated by unanticipated risk. Bhagat and Bolton (2014) state that the executive compensation of a bank should only consist of restricted stock and restricted stock options. ‘Restricted’ means that the CEO cannot exercise its options or sell its shares for two to four years after an executive stops at the particular institutions (Bhagat and Bolton, 2014).

2.5 Hypotheses

At the end of this research, the exact relationship between firm performance and the compensation scheme of an executive should be analyzed. A compensation scheme of a CEO is build out of several components, which are in this thesis used as the explanatory variables. Based on former research, these components and its theoretical effect on the firm performance were described in the literature review. Thus, based on this literature review, it is possible to make a few hypotheses about the results of the research analyze. There will be made four hypotheses predicting the effect of the coefficients of several explanatory variables.

In 1995 Mehran analyzed this relationship as well and his results gave empirical evidence that the performance of a firm has a positive relation with the percentage of the compensation scheme that is equity-‐based. In this research two forms of equity-‐ based compensation are used within the regression model: stock-‐awards and option-‐ awards. Also according to Hall and Liebman (1998), the change in value of the

executives holdings in stock options and general stock, is the base of the generated relationship between firm performance and CEO compensation. The findings of these two researches leads to the following two hypotheses:

(H1): the coefficient of the explanatory variable ‘stock awards’ will be positively correlated with the dependent variable ‘firm performance’.

(15)

(H2): the coefficient of the explanatory variable ‘option awards’ will be positively correlated with the dependent variable ‘firm performance’

Thus, the expectation may be that the components stock awards/option awards, as a percentage of the total compensation, have a positive effect on the performance of a firm.

Some researchers merge stock and option awards as one variable; equity based awards. This can thus be seen as one equal component of the CEO compensation package instead of two different components. This leads to the following hypothesis.

(H3): the coefficient of the explanotary variable ‘equity based awards’ will be positively correlated with the dependent variable ‘firm performance’

According to Fahlenbrach & Stulz (2011), CEO’s took risks before the crisis and whenever an executive takes a risk that is not aligned with the interest of the shareholder, this executive should sell some shares. Thus, it did not have a positive effect if CEO’s got awarded in shares during the credit crisis (Fahlenbrach & Stulz, 2011). This leads to the fourth hypothesis:

(H4): the coefficient of the explanotary variable ‘stock awards’ will be negatively correlated with the dependent variable ‘firm performance’ in case of a credit crisis

All four hypotheses are based on equity-‐based compensation components, because, based on the literature review, these are the most important and effective components of a compensation package.

(16)

3. Data & Methodology

3.1 Data

To do an analysis and test the hypotheses made to answer the research question, certain data streams were used. This paper focuses on a sample of 154 commercial banks from the United States, because of the specification in the US banking industry. This sample of commercial US banks used, is one of the factors that makes this research different from other studies. The data needed to measure the used components of an executive

compensation scheme, was obtained from the Compustat database of WRDS. The

components of compensation that were taken from this database are: 1) Salary 2) Bonus 3) Stock Awards 4) Option Awards 5) Other Compensation and 6) Total compensation. Further, 7) the Company Name, 8) Company ID Number and 9) the Company CEO Name of each bank, were retrieved from the Compustat database. In contrast with former research, the ‘Total compensation’ is not used as an independent variable. In this paper ‘Total compensation’ is only used to determine the other components as a percentage of this total compensation. All this compensation and company data was received from the time period of 2005 until 2016.

Because this paper researches the relationship between the executive

compensation structure and the firm performance, financial data containing this firm performance of each bank was obtained from the Capital-‐IQ database of WRDS. The data obtained was also from the time period 2005 until 2016 and from the same sample of 154 banks. Firm performance is measured with the Tobin’s q and as already mentioned in the literature review, Tobin’s q is defined as: 𝑠ℎ𝑎𝑟𝑒 𝑝𝑟𝑖𝑐𝑒 × 𝑠ℎ𝑎𝑟𝑒𝑠 𝑜𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔 + 𝑓𝑖𝑟𝑚!_{𝑠𝑜𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔 𝑝𝑟𝑒𝑓𝑒𝑟𝑟𝑒𝑑 𝑠𝑡𝑜𝑐𝑘 + 𝑑𝑒𝑏𝑡 ÷ 𝐵𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠}

Consequently, the 10) Share Price, 11) Number of Shares Outstanding, 12) Outstanding Preferred Stock and 13) Debt for each of the 154 firms were taken from the database. All the values of the data retrieved, were denoted in US dollars. The control variable used in the model is the Firm Size, which is equal to the Number of Shares Outstanding times the Price per Share. The data of these components were already received by obtaining the data needed to measure the Tobin’s q.

(17)

3.2 Model and regression

This research paper can be defined as a quantitative analysis. In short, with a

quantitative analysis, one uses numbers to resolve and describe a research problem. This can be achieved by collecting data, measuring data, summarizing data and drawing conclusions from the data. Advantages of a quantitative analysis may be that the data obtained, can be generalized, results are more accurate and results obtain a greater number of subjects. Accuracy of the results also depends on the size of the dataset, which is the reason why this study used a great data set. In this model, panel data (or cross sectional time-‐series data) is used to perform the regression and to, eventually, draw a conclusion from this regression. Panel data is a style of data on an economic variable that incorporates multiple economic units as well as multiple time periods, hence illustrate both cross sectional variation and time series variation. Since this thesis also uses a multiple time period (ten different years) and multiple economic units

(different components of a CEO compensation scheme) and wants to make a comparison over time, panel data should be suitable. With this panel data the behavior of the

performance of banks from the sample is observed over a time period of ten years (2007-‐2016). To identify the behavior of every bank’s performance, each Company-‐CEO combination in the data set is observed at ten points of time. Panel data is used because the firm performance and the CEO compensation package changes over time and not across banks. With a Panel data set, indexing observations starts with _{𝑡
as
well
as
𝑖
to} analyze between the observations of Company-‐CEO combination _{𝑖

at
different
points
of} time _𝑡.

When regressing panel data, there may be some complications with respect to the expected correlation between the components of the CEO compensation package and the firm performance. One explanation is that there might be individual differences across US banks that cannot be explained by CEO compensation schemes. This may be explained by firm specific characteristics, which can be seen as a constant effect over the years, which differs at each bank. This time invariant bank specific factors can also be seen as cross sectional differences. Similarly, time-‐specific characteristics, like an

(18)

are used when performing a panel data regression. In this thesis two fixed variables are used, one for the cross sectional differences and one for the time-‐series variation. One multiple regression model was concluded, relying on these assumptions for panel data:

𝑌𝑖𝑡=𝛼𝑖+𝑡𝑡+𝛽0 +𝛽1𝐶𝑜𝑚𝑝𝑖𝑡+𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡+𝛽3𝐶𝑜𝑚𝑝𝑖𝑡∗𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ln(Firm Size)+𝜀𝑖𝑡

Where:

𝑌𝑖𝑡: the Tobin’s q for a company-‐CEO combination 𝑖 in year 𝑡. 𝛼𝑖: Company-‐CEO combination fixed effect

𝑡𝑡: the time fixed effect β0: constant coefficient

𝐶𝑜𝑚𝑝𝑖𝑡: compensation variable for company-‐CEO combination 𝑖 in year 𝑡

𝛽1: the effect of 𝐶𝑜𝑚𝑝𝑖𝑡 on Tobin’s q (A 1%-‐point increase in Comp leads to a _{𝛽1*0.01
increase
in
Tobin’s
q)}

𝐶𝑟𝑖𝑠𝑖𝑠𝑡: dummy that is 1 for crisis years 2007 and 2008

𝛽2: the difference in Tobin’s q in crisis years 2007 and 2008 𝛽3: the effect of compensation in crisis years 2007 and 2008

ln(Firm Size): the natural logarithm of the size of the firm (market capitalization: share price*shares outstanding). This is a control variable.

𝜀𝑖𝑡: error term

The model above is a theoretical model and the variable _{𝐶𝑜𝑚𝑝𝑖𝑡
can
be
defined
in} different ways. The aim of this paper is to analyze the effect of five particular

components of the CEO compensation scheme on firm performance and therefore the following five different definitions of _{𝐶𝑜𝑚𝑝𝑖𝑡
are
used:
salary,
bonus,
stock
awards,} option awards and other compensation. The effect of each of the components on the firm

(19)

performance will be studied by doing an analysis. After regressing the separate and unique combination of components

First the salary component will be analyzed and the following multiple regression model is used when doing this.

(1) _{𝑌𝑖𝑡=𝛼𝑖+𝑡𝑡+𝛽0 +𝛽Salary𝑖𝑡+𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡+𝛽3Salary𝑖𝑡∗𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ln(Firm
Size)+𝜀𝑖𝑡}

Salary is seen as the most fixed component of a CEO compensation package (Tosi et al, 1989). As already mentioned in the literature review, there has been done a lot of research to the effect of base salary on firm performance and whether it has to be a big or small part of the CEO compensation package. With this model the coefficient will be determined and tested to establish whether base salary has an effect on the

performance of banks in the US. Furthermore, both the effect in the crisis years and the effect in the years after the crisis will be determined using _{𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡 as a dummy (1 for} years 2007 and 2008, 0 for all other years). Each salary is taken as a percentage of the appropriate total compensation (𝑆𝑎𝑙𝑎𝑟𝑦 ÷ 𝑇𝑜𝑡𝑎𝑙 𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛). With the total

compensation the following is meant: salary + bonus + value of stock awards + value of option awards + other compensation.

The bonus is the second component of the CEO compensation package that will be analyzed. The definitions used in bonus schemes always differ between companies; a firm can award a fixed cash bonus, a bonus per unit of output etc. According to Murphy and Jensen (1990) bonus is a big part of the Compensation package. To analyze the effect on Tobin’s q of this bonus, the following model is used.

(2) _{𝑌𝑖𝑡=𝛼𝑖+𝑡𝑡+𝛽0 +𝛽Bonus𝑖𝑡+𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡+𝛽3Bonus𝑖𝑡∗𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ln(Firm
Size)+𝜀𝑖𝑡}

Bonus is in this model also taken as a percentage of the total compensation package. Both the effect in the crisis years and the effect in the years after the crisis will be analyzed again.

(20)

The third and fourth components used in this thesis are stock and option awards. With stock and option awards an executive’s pay is based on the equity of the firm (Mehran, 1995). This will lead to the following two models.

(3) _{𝑌𝑖𝑡=𝛼𝑖+𝑡𝑡+𝛽0 +𝛽1Stock_awards𝑖𝑡+𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡+𝛽3Stock_awards𝑖𝑡∗𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ln(Firm} Size)+_𝜀𝑖𝑡

(4) 𝑌𝑖𝑡=𝛼𝑖+𝑡𝑡+𝛽0 +𝛽1Option_awards𝑖𝑡+𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡+𝛽3Option_awards𝑖𝑡∗𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ln(Firm Size)+_𝜀𝑖𝑡

These models analyze the effect of stock and option awards on the performance of the banks from the sample. The crisis dummy, distinguishes the effect in the years of the crisis from the effect in the years after the crisis. These variables are also taken as a percentage of the total compensation. Some researchers argue that stock and option awards can be seen as one component of the CEO compensation package; equity based-‐ awards. To analyze the effect of both components together, the following formula is used. (5) 𝑌𝑖𝑡=𝛼𝑖+𝑡𝑡+𝛽0 +𝛽1Equitybased_awards𝑖𝑡+𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡+𝛽3Equitybased_awards𝑖𝑡∗𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ln( Firm Size)+_𝜀𝑖𝑡

The last component that is used to measure the compensation package of a CEO is ‘other compensation’. Other compensation can be described as things like insurances,

vacations, Christmas gifts etc. To analyze the effect of this last component on the firm performance, the following model is used.

(6) 𝑌𝑖𝑡=𝛼𝑖+𝑡𝑡+𝛽0 +𝛽1Other_comp𝑖𝑡+𝛽2𝐶𝑟𝑖𝑠𝑖𝑠𝑡+𝛽3Other_comp𝑖𝑡∗𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ln(Firm Size)+_𝜀𝑖𝑡

(21)

3.3 Descriptive statistics

In table 1, the descriptive statistics of all the various independent variables are given. The amount of observations is very high, because it contains all different amounts of a particular variable from each US bank from the dataset and from each year of the time period used (2005-‐2012). Every independent variable in table 1 is described as a percentage of the total compensation package, except for the variable ‘Total_c’. This variable can be described as the absolute value of the total compensation. It is odd that the minimum value of the total compensation is 0, but an explanation has been found for this. J.S Sidhu has been the CEO for 9 years at Customers Banccorp Inc and earned zero compensation the first year he became CEO, which was in 2008. This zero compensation may be the case, because 2008 was one of the years that the crisis was at its peak and thus the bank may have faced such amounts of debt that it could not afford to

compensate its staff. Nevertheless, 2008 was the only year that J.S Sidhu received zero compensation, thus the expectation is that it would be an outlier. However, because of a standard deviation of $2536.57, the minimum value of zero is theoretically not an outlier. The maximum of $32974.16, on the other hand, can be described as an outlier. This outlier was not removed from the sample because the year beforehand G. Kennedy Thompson, CEO of Wachovia Corporation, earned $16333.29, which is approximately the half of $32974.16. The table demonstrates that all minimum and maximum values differs widely for each variable. This can be the case, because every US bank experienced this period of time, with a lot of economic shocks and fluctuations, in its own way.

Table 1: descriptive statistics

Variable Observations Mean Std. Dev. Min. Max. Salary 6439 .453 .228 0 1 Bonus 6439 .0694 .131 0 .976 Stock_a 6439 .240 .589 -‐1.948 41.903 Option_a 6439 .079 .277 -‐1.474 16.822 Equity_a 6439 .320 .803 -‐3.378 58.725 Other_c 6439 .083 .127 0 1 Total_c 6439 1737.19 2536.57 0 32974.12

What is the relationship between the compensation package of a CEO and the firma performance within the US banking industry? : a comparison between the financial crisis and the period after