Frederic Brault & Arjen Hendrick Sanderman
The behavioral agency theory: a general approach towards a contingent understanding of the
pay-performance relationship
Business Administration Master’s Thesis
30 ECTS
Term: Spring 2016
Supervisor: Hans Lindkvist
3
i. Abstract
In this thesis the pay performance relationship is discussed within the field of executive remuneration. A topic that received a lot of attention during the financial crisis and times of economic recession. The starting point of this is a broken pay setting process, where existing neoclassical theory is highly divided in providing explanations for the pay performance relationship. A rather new theory, called the behavioral agency theory, provides a new perspective on the pay performance debate and argues for a rather contingent approach. In this paper the selected underlying assumptions of the behavioral agency theory were tested within a European context. The index used to select the companies is the Eurostoxx50 index.
The underlying assumptions are the effects of loss averse behavior by principals,
agent motivation and agent time preferences. Significant results were found in
supporting the loss averse behavior by principals and Agent motivation. The agent
time preferences could not be explained by the selected method.
4
ii. Table of Contents
I. ABSTRACT ... 3
II. TABLE OF CONTENTS... 4
III. WORK DISTRIBUTION BETWEEN GROUP MEMBERS ... 5
IV. LIST OF ABBREVIATIONS AND IMPORTANT CONCEPTS ... 6
1. INTRODUCTION ... 7
1.1 THEORETICAL BACKGROUND ... 7
1.2 AIM OF THE RESEARCH ... 9
2. THEORETICAL FRAMEWORK ... 10
2.1 THE AGENCY THEORY AS A FOUNDATION FOR ORGANIZATIONAL PERFORMANCE ... 10
2.2 TOWARDS A BEHAVIORAL AGENCY THEORY ... 12
2.3 PRINCIPAL LOSS AVERSION ... 12
2.4 MOTIVATION ... 14
2.5 AGENT TIME-PREFERENCES ... 15
2.6 THE MEASURES OF ORGANIZATIONAL PERFORMANCE... 16
2.7 SUMMARY OF THE CHAPTER ... 17
3. METHOD ... 19
3.1 UNDERLYING CASE ... 19
3.2 SAMPLE DESCRIPTION ... 19
3.3 DATA COLLECTION PROCEDURE ... 21
3.4 RESEARCH DESIGN ... 21
3.5 METHODOLOGY ... 22
3.5.1 The construct of the hypotheses ... 22
3.5.2 Measures of organizational performance ... 23
3.5.3 Test of the hypotheses ... 24
4. EMPIRICAL FINDINGS AND RESULTS ... 25
5. DISCUSSION AND ANALYSIS ... 30
5.1 THEORETICAL DISCUSSION AND ANALYSIS ... 30
5.2 DISCUSSION OF RESEARCH DESIGN ... 33
5.3 MANAGERIAL IMPLICATIONS ... 36
5.4 LIMITATIONS AND FUTURE RESEARCH ... 37
6. CONCLUSION ... 39
REFERENCE LIST ... 40
5
iii. Work distribution between group members
Chapter/part of the thesis Author
I. ABSTRACT Arjen Hendrick
II. TABLE OF CONTENTS Arjen Hendrick
III. WORK DISTRIBUTION BETWEEN GROUP MEMBERS Arjen Hendrick
IV. LIST OF ABBREVIATIONS AND IMPORTANT
CONCEPTS Arjen Hendrick
1. INTRODUCTION Arjen Hendrick
1.1 THEORETICAL BACKGROUND Arjen Hendrick
1.2 AIM OF THE RESEARCH Arjen Hendrick
2. THEORETICAL FRAMEWORK Arjen Hendrick
2.1 THE AGENCY THEORY AS A FOUNDATION FOR
ORGANIZATIONAL PERFORMANCE Arjen Hendrick
2.2 TOWARDS A BEHAVIORAL AGENCY THEORY Arjen Hendrick
2.3 PRINCIPAL LOSS AVERSION Arjen Hendrick
2.4 MOTIVATION Arjen Hendrick
2.5 AGENT TIME-PREFERENCES Arjen Hendrick
2.6 THE MEASURES OF ORGANIZATIONAL
PERFORMANCE Arjen Hendrick
2.7 SUMMARY OF THE CHAPTER Arjen Hendrick
3.1 UNDERLYING CASE Arjen Hendrick +
Frederic
3.2 SAMPLE DESCRIPTION Arjen Hendrick +
Frederic
3.3 DATA COLLECTION PROCEDURE Arjen Hendrick +
Frederic
3.4 RESEARCH DESIGN Arjen Hendrick +
Frederic
3.5 METHODOLOGY Arjen Hendrick +
Frederic
4. EMPIRICAL FINDINGS AND RESULTS Arjen Hendrick +
Frederic
5.1 THEORETICAL DISCUSSION AND ANALYSIS Arjen Hendrick
5.2 DISCUSSION OF RESEARCH DESIGN Arjen Hendrick +
Frederic
5.3 MANAGERIAL IMPLICATIONS Arjen Hendrick
5.4 LIMITATIONS AND FUTURE RESEARCH Arjen Hendrick +
Frederic
6. CONCLUSION Arjen Hendrick
DATA COLLECTION REMUNERATION Frederic
DATA COLLECTION ORGANIZATIONAL
PERFORMANCE Arjen Hendrick
6
iv. List of abbreviations and important concepts
Abbreviations
CEO : Chief Executive Officer
ES50 : Eurostoxx 50
ES600: Eurostoxx 600
EU: European Union
ROA : Return on Assets
U.K.: United Kingdom
U.S. : United States
Important concepts
Agency theory a theory that applies for publically owned companies where the owner(s) and the executives are not the same person. The agency theories argues for interest alignment to solve this moral hazard problem. (Ross 1973)
Behavioral agency theory a variant of the agency theory that acknowledges the foundation of the agency theory but argues for behavioral underlying assumptions (Pepper & Gore 2015)
Managerial power theory a theory that argues that more power of the CEO leads to a higher remuneration and a weaker pay performance relationship (Tosi et al. 2000)
Pay performance relationship the relationship between the salary of the CEO (in this study) and the performance of an organization/firm Remuneration the synonym of salary or wages
7
1. Introduction
1.1 Theoretical background
Remuneration and how to establish it has been studied for many years. Interest in the topic grew in and after the year of the financial crisis in 2008 when bonuses reached a new all- time high in the years before. Jensen et al. (2004) found that the average executive compensation increased from $85.000 dollars in 1970 to $14 million in 2000 before declining to 9.4 million in 2002. After 2002 it started growing again until the aforementioned financial crisis in 2008. The crisis made topics as fairness and risk more important and questions whether the pay performance relationship is still intact. Until today there are two theories that are laying the foundation of the remuneration literature in providing a general explanation for executive remuneration. The first significant theory looks upon executive remuneration as a performance contract between the owner (principal) and the executive (agent) called the Agency theory (Ross 1973). The Second theory finds its nature in aspects executive behavior and argues that the power of the manager determines its total remuneration (Bebchuk & Fried 2004). The more power the manager has over the pay setting process, the higher his own remuneration will be. This theory is called the managerial power theory. To develop an understanding of the ongoing remuneration debate an overview of both theories
Within economics and finance, the main theory is still a neoclassical theory that is often used in market based explanations (van Essen et al. 2015). This theory outlines that the height and evolvement of bonuses can be explained along organizational performance, serving as a performance contract (Ross 1973). From an agency perspective, the remuneration should serve to motivate the board of directors and therefore align the interest of the company owners with its directors. This theory finds its origin in a study conducted by Berle and Means (1932). They recognized that the separation between directors and owners caused certain problems that needed to be solved. One of the problems that they addressed is the moral hazard problem. Since the agent does not use his own money but the money of the principal, he will spend it differently as if it was his own. The agent is more interested in maximizing his own wealth and therefore a conflict arises with the principal.
A principal must therefore implement systems that encourage the executives to act in his interest. According to the agency theory, the principal (shareholder) may establish a remuneration agreement using observable performance measures, prompting the agent (the executives) to act in the interests of the principal (Davis et al. 1997).
8
Scholars in favor of the managerial power theory often refer to a well-known meta-analysis in executive compensation conducted by Tosi et al. (2000). They combined the results and analyzed 42 studies within United States (U.S.) companies. The results showed that the variance of executive remuneration can be mostly determined by company size (40%) rather than organizational performance (<5%). This study highlighted the complexities within the pay-performance debate. Therefore it is often used to strengthen the managerial power theory (Bebchuk and Fried 2004). Another influential meta-analysis compared chief executive officer (CEO) total compensation versus performance pay-sensitivity and conclude that the managerial power theory indicates that it is suitable for predicting compensation variables as total cash payments and total remuneration but weaker results for the pay-performance relationship. An interesting findings concerning this pay- performance relationship is that when the manager is ought to have more influence on the pay performance process there is a stronger pay performance relationship measured. Also board independence and institutional ownership positively moderate the pay-performance relationship (van Essen et al. 2015).An often overlooked point is the correlation between organizational size and executive remuneration. Which implies rather a strong relationship between organizational size over the concept of managerial power. Therefore, regardless of the interesting conceptualization towards the direction of the managerial power theory both studies, and accordingly most studies, show the most significant findings in relation to company size and company performance. The concern of Gregory-Smith (2012) that the pay setting process is broken, because of their foundation based on orthodox economic theories, seems right. These basic orthodox or neoclassical theories provide inconsistent proof and fail in providing a one- size fits all solutions. However, by discarding the agency theory prematurely they might miss out on the opportunity of understanding the incentive alignment construct (Nyberg et al. 2010). Their arguments for weak findings were the inconsistent measures (Nyberg et al.
2010).
Based on our findings in the literature it can be said that prior research is highly divided when it comes to executive remuneration and the adoption of these two theories. The agency theory seems to justify high remuneration because of a performance contract, where the managerial power theory argues for a dysfunctional system where executives reward themselves more than is necessary. However, consistent proof of both theories is still lacking. Instead of providing more mixed results in both constructs this thesis aims for
9
explaining the pay performance relationship by applying a contingent framework. This contingent framework uses the principles of the agency theory and build further on behavioral factors often used by managerial power theorists. This framework is called the behavioral agency theory (Pepper & Gore 2015). In the literature review behavioral factors are used to test a modified agency theory. These propositions form the foundation for this study and will be tested among the top European companies listed in the Eurostoxx50 (ES50) index.1.2 Aim of the research
Building further on the problematization of the current research within studies about the pay performance relationship. The conclusion is drawn that based on neoclassical theory consistent results lack in supporting a clear answer to prove a pay performance relationship.
In the theoretical framework an alternative approach is developed. Three important variables that differ from the neo-classical agency theory were formed based upon the behavioral agency theory. This theory uses behavioral economics as a foundation to explain variations within the pay performance relationship. This theory is mainly derived from a paper about this subject written by pepper and Gore (2015). In their paper an inductive approach is used to explain this construct. In order to create a general understanding of the theory this paper uses a deductive approach to explain behavioral agency tendencies within the pay performance relationship. The aim of this paper is to theorize, explain and support the underlying assumptions of the behavioral agency theory and therewith explain contingent behavior within the pay performance relationship.
Thereby a sample consisting of companies within Europe with the largest market capitalization is used. Most studies within the executive remuneration area and about the pay performance relationship are performed with U.S. data. Besides that it is not the main aim to perform a country/continent comparison, the use of a European sample might already lead to interesting findings within the pay performance debate.
In the next chapter the theoretical framework of this thesis will be formed, starting with the agency theory and its principles. Consecutively the behavioral agency theory is explained and the focus fundamentals of this thesis are elaborated on.
10
2.
Theoretical framework
The theoretical framework forms the foundation of this study and argues from a theoretical perspective for the hypotheses of this study. As elaborated on in the introduction, neoclassical theories do not provide a consistent answer in explaining variations in executive remuneration. The introduction ended with the behavioral agency theory as it is constructed by Pepper and Gore (2015). This theory provides a new dimension of how executive remuneration should be looked upon and challenges the underlying assumptions of the agency theory. Figure 1 is a reflection of the theoretical framework. The numbers refer to the chapters the particular theory is discussed in. The first chapter starts by explaining the agency theory. After laying out the principles of the agency theory a connection to the behavioral agency theory is made. Consecutively, the specific components of the behavioral agency theory are reflected upon. These are behavior, motivation and agent time preferences. The theoretical explanation of organizational performance factors follows. The chapter ends with a summary of the hypotheses derived from the theoretical framework.
Fig 1: The conceptual model of the theoretical framework.
2.1 The agency theory as a foundation for organizational performance
As mentioned in the introduction of this paper, the agency theory is often used in the studies of executive remuneration. One of the first papers that addressed the agency theory and gave a solution to aligning the interests of shareholders and executives was written by Berle and Means (1936). The basic notion was that the change of the relation between ownership and leading the company leads to a situation with principle (the owner) and an agent (the executives). In executive remuneration literature the agency theory is frequently used
11
because of the fact that from an agency point of view, the executive compensation can be seen as a contract between the agent and the principal (shareholder). This is done in order to align the interest of the shareholder (principal) with the interest of the executive director (agent) and can be seen as a justification for the remuneration paid to the executive director (Westphal & Zajac 1998). The salary of the director can be separated in three different types of costs according to the agency theory. These three types are monitoring expenditures by the principals, the bonding expenditures by the agent and the residual losses. The monitoring expenditures as cost related to the monitoring process of the agent by setting out performance indicators as guidance. In addition some costs may occur for resources at the expense of the principal that let the agent act in a favorable way from a shareholder point of view. Because the agent cannot make the right decision in all situations, the residual costs are the costs as a consequence of those “poor” decisions (Jensen & Meckling, 1976). The agency theory is founded on the presumptions that the agent normally would act in his own interests rather than in favor of the shareholder because he is rent seeking for his own purposes. The agent can be financially stimulated and is contingent on effort. Furthermore he makes rational decisions, is risk averse and preferences time according to the exponential discount function (Jensen 1998).In the remuneration debate the agency theory has shown mixed results in proving strong evidence for this theory. Sub categories of the agency theory and executive remuneration studies are the pay performance relationship and agent and shareholder wealth alignment.
In the early nineteen nighties an important shareholder wealth alignment study investigated the relationship between share price and CEO pay and found no supporting results (Jensen and Murphey 1990). This in contradiction with a later study that tested shareholder and CEO interest alignment (Nyberg et al. 2010). Possible reasons besides the different approaches could be for example that between the times of both studies there were changes in the structure of the remuneration packages. In the years after the publication of Jensen and Murphy (1990) long term performance packages became popular which clearly improved the relationship between shareholder and CEO wealth alignment.
The first significant meta-analysis found stronger results between company size and only a weak relationship with company performance (Tosi et al. 2000). Based on these findings one of the first studies considering this topic of managerial power theory was performed by Bebchuk and Fried (2002). The shift towards a managerial power theory did neither show consistent results. An extensive study performed from 1936 to 2005 proved that neither the agency theory nor the managerial power theory is a fully consistent theory
12
(Frydman & Jenter 2010). Twelve years later a new and bigger meta-analysis was performed. In this analysis company performance had a more significant value, but company size remained to be the biggest influencing factor (van Essen et al. 2012). As for example Gregory-Smith (2012) argued for a broken pay setting process.Seen as the opposite of the Agency theory is the Stewardship theory. Doucouliagos (1994) proposed another approach of human behavior, building on the work in the fields of psychology and sociology. In fact, selfishness is not sole motivation of human action.
Nevertheless, there is not necessarily a divergence of interests between principal and agent (Donaldson and Davis, 1991). These assumptions about human motivation, prompted the authors to suggest assumptions of the stewardship theory and consider the agency principal of motivation differently (Donaldson, 1990). The stewardship theory assumes that executives are intrinsically motivated and can considered as being "stewards" of their company. In the stewardship theory, there is not always a divergence of interests between executives and shareholders. Leaders are satisfied when they are able to improve the performance of their organization and to care about their personal interests. In the stewardship theory, leaders compromise between the interests of their organization and their personal interests (Wasserman, 2006). This theory contradicts the need of expensive agency costs and the use of bonuses to resolve the agency problem. Although this theory does not provide a constructive solution or explanation for the level of remuneration, it indicates that some of the foundations of the agency theory might be more complicated as argued for.
2.2 Towards a behavioral agency theory
A different dimension is provided by a rather new theory called the behavioral agency theory, this theory takes the flaws of the agency theory into account. This theory recognizes fundamentals of the agency but argues for new approaches based on a different and more current research perspective. The behavioral agency theory looks for explanations within the subject of behavioral economics. In this thesis factors of the behavioral agency theory concerning the subjects of principal risk preference, motivation, agent time-preference, inequity aversion and goal setting are used to explain the construct.
2.3 Principal loss aversion
In agency theory, the agents and principals have a neutral stance on risk preference and act in a linear way to organizational performance and executive pay. Meaning that their response will be different in a situation when they are losing capital other than when they
13
are in a situation where they are gaining. This is based upon the notion that executive remuneration is solely to motivate executives to maximize shareholder value (Jensen &Murphy 1990). This rather high cost is based upon the information asymmetry between the principal and the agent and therefore needs to be aligned (Gomez-Mejia 1998). Thus, executive remuneration could be seen as a second best solution for the principal. The principal rather has its interest aligned at the lowest costs possible, but needs to pay for performance alignment (Krause et al. 2014). The task of the principal is not to formulate the executives’ remuneration plan, but only to approve or disapprove. Furthermore, the principals need to assess whether the agents earned their remuneration or not (Fama 1980).
From a rational perspective this seems logical. Eisenhard (1989) compares it with the product/quality paradigm. As a customer you want to pay the lowest price possible for a certain service and as an investor you want to pay the lowest price possible for the services of the CEO.
However, a handful of scholars have disproven this fact of rational behavior and acknowledged the differences between rational behavior and actual behavior (Tversky &
Simonson, 1993; Krause et al. 2014; Pepper & Gore 2015). The behavioral agency theory suggests an alternative view and argues for different behavior in a losing position. This argument is formulated based on the prospect theory, which initially stated that a person’s reference level alters how decisions are made (Pepper & Gore 2015). Later on Kahneman
& Tversky (1991) found that also in a neutral position preferences were based upon this reference level. This theory provides both a different perspective on risk preferences among shareholders and also argues for limited rationality. Another important finding, based on the prospect theory is the fact that people are afraid of losses and are therefore more likely to take risks in a loss situation rather than take risks to gain (Tversky & Kaneman 1981).
Compared to the agency theory Pepper and Gore (2015) argue for a contextual approach where the perceived loss position leads to risk enhancing decisions. An experimental study performed by Krause et al. (2014) argues for loss-averse behavior by principals and relates the case to shareholder voting on executive pay as displayed in figure 1.
14
Fig 2: Agency costs and shareholder decision frames based on (Krause et al. 2014)
The participants of their experiment acted in a negative different way when they were losing capital. Besides this loss averse behavior another point they found was that principals do favor strong performance but not necessarily prefer low CEO rewards. This dimension provides a different view of how to look upon the principal’s decision making when it comes to remuneration. Based on this prospect theory we argue for a weaker pay performance relationship in shareholder loss situations. This is based upon the fact that shareholders are more likely to vote against new remuneration packages, despite of an increased pay performance relationship.
2.4 Motivation
In the agency theory the general notion is that the agent is solely motivated to gain personally and can be financially stimulated in order to align interests. Even though this seems logical from a rational perspective, behavioral economic theory argues for a more complex understanding of human behavior. This theory acknowledges the fact that the purpose of incentives is to motivate the agent as for example argued for by Jensen and Murphy (1990), but rather focuses on the extend the agent can be, and is, motivated because of these measures. Concerning the topic of agent-motivation there is not only financial stimulation that affects behavior. One of the early studies regarding this topic found that there is a distinction between intrinsic and extrinsic motivation (Deci & Ryan 1985).
Intrinsic motivation is defined as the motivation coming from other factors than a financial reward, looked upon as the agent performing a task for his or her own satisfaction. Extrinsic motivation is caused solely because of instrumental value. By relating the topic to the pay performance relationship it could be argued for that the relationship could be weaker because the motivation of the CEO is not solely extrinsically. Deci and Ryan (1985) argue that after a certain point monetary compensation can reduce the intrinsic motivation, and that therefore too much financial stimulation works counterproductive. Later studies criticized design of the incentive system is poorly executed and argued for a “crowding
Neutral position
Gain position
Loss position
Neutral position High
CEO rewards Low CEO rewards
Strong organizational performance
Poor organizational
performance
15
out” effect within executive remuneration that influences the relationship between intrinsic and extrinsic motivation (Frey & Jegen 2001; Sliwka 2007). Despite of their criticism, the scholars recognized the problem of applying measures to develop a general econometrical theory (Frey & Jegen 2001). As a general understanding was tried to be found, answers in specific parts of were analyzed. Concerning the measure of stock based compensation it is found that it sometimes can cause unfavorable decision making (Devers et al. 2008). This finding is a good example that contingent situations are likely to apply. An interesting distinction is made between motivation and interest alignment by Pepper & Gore (2015).They argue for emphasizing the interest alignment over motivating the agent. In their proposition they define the intrinsic and extrinsic motivation as measurable factors based on contingent factors related to agent’s situation. However, from an investor or outsider’s perspective where information asymmetry exists (Berle and Means 1936), it is not possible to collect this data and therefore create a general theory upon. In order to generalize this statement Murdock (2002) argued that the intrinsic motivation should be seen rather as an addition other than a moderation.
2.5 Agent time-preferences
In the Agency theory the time-preference of the agent is looked upon as in many financial models, which is done by discounting the value of money exponentially. This is based upon the assumption that instant reward is the most preferred option, the further away the reward the less preferred by the agent. This preference is equally distributed over time (Ross 1973).
The behavioral agency theory looks upon it from a behavioral economic perspective and rather argues for a hyperbolic discount function. The biggest difference is the time perception. In the behavioral agency theory the time preference is distributed hyperbolically instead of equally. This means that rewards in the nearby future more rapidly lose their value to the agent compared to the agency theory. One of the first studies that introduced hyperbolic time discounting is Ainslie (1991), in an experimental study. Graves and Ringuest (2012) argue for a general acceptance of this theory within behavioral economics. Based on this theory the agent is more likely to make decisions that maximize short term gains and avoid long term plans as in the agency theory. Concerning the topic of remuneration it is more likely for the agent to aim for either the short term fixed remuneration or short term bonuses. It can be argued for that long term stock option plans are therefore less valuable for the agent or become more relevant in times of granting.
Although both theories discount the value of money over time, from an agency perspective long term share- and option plans are one of the most measurable forms of interest alignment. Nyberg et al. (2010) found a more significant relationship between CEO return
16
and interest alignment than previously argued for. From a behavioral economic perspective the relevancy seems lower. Another notion to take into account is that there are other factors than organizational performance influencing the share price and therefore affecting long term option- or share plans. According to Richard et al. (2009) stock market performance is not solely influenced by organizational performance but is also affected by the volatility of markets, the economy and psychological factors. Hamann et al. (2013) argue that environmental instability should be avoided in non-longitudinal studies measuring organizational performance. By both taking into account the behavioral economical reasoning and organizational stability measures the years of high stock market volatility are expected to be weaker.2.6 The measures of organizational performance
In the literature there seems to be no consensus on valid measures that explain organizational performance. In a construct validity Hamann et al. (2013) investigated based on a factor analysis which indicators should be used in explaining organizational performance and how these can be classified. Based on previous research four different models were created and tested. In the study of Hamann et al. (2013), three limitations of previously conducted research is pointed out. The first point focuses on the number of different dimensions of organizational performance. In early studies the discussion whether to use three or four different dimensions seemed already an argument (Fryxell & Barton 1990; Venkatraman & Ramanujam 1987). It is argued for that the fewer indicators to explain a construct make it harder to justify a valid explanation (Hamann et al. 2013).
Second, the argument is the choice of the number of indicators used to measure organizational performance. The reasoning behind the choice of organizational performance measures refers often to the use in previous papers, as for example Murphey et al. (1996). Some of those measures are clearly not related to organizational performance, as for example size and static balance sheet measures. Studies with significant impact of scholars in favor of the managerial power theory also used various indicators of organizational performance. The meta-analysis performed by van Essen et al. (2015) used any indicator of financial performance which is either a market based- or an accounting performance measure. Also the often referred to study of Tosi et al (2000) looks at organizational performance in a wide range and does not specifically uses one construct to define it. Although the sample size of those studies is significant, the different measures could give inconsistent results.
17
Third and last, cash flow measures are almost always overlooked when it comes to organizational performance. This extra dimension is important since it limits the view on organizational performance and the effects of it. The dimension of accounting measures is in most studies looked upon as one. After taking into account these limitations a factor analysis is performed to construct an optimal model that reflects organizational performance (Hamann et al. 2013). As a result a four dimensional model of performance measures is recommended. This model separates accounting measures into profitability measures and liquidity measures and consists of organizational growth and stock market performance. In contrast to the aforementioned profitability and liquidity measures, the stock market performance focuses on future instead of the past. By applying these performance measures a clear reflection of organizational performance is created and should reflect the construct of organizational performance at best (Hamann et al. 2013).2.7 Summary of the chapter
This chapter starts with a theoretical discussion of the agency theory and its principles and builder further on the principles of the behavioral agency theory. The three hypotheses constructed from the theoretical framework are based upon principal risk aversion, agent motivation and agent time preference.
Within the agency theory the agent and the principal are able to make rational decisions in every situation regardless of situational factors, the behavioral agency theory distinguishes different behavior in loss, neutral and gaining situations. The strongest effects were found in situations of loss situations. This study connects the effect of principals in loss situations, also called loss aversion, with the topic of executive remuneration. Based on this behavior the following hypothesis is constructed:
Hypothesis 1: In years of shareholder losses the link between shareholder remuneration and organizational performance is expected to be weaker in the year after the shareholders loss.
Within the topic of agent motivation the agency theory argues for solely extrinsic motivation based on monetary reward. The underlying assumption is that the agent is solely interested in maximizing his or her own personal gains. The behavioral agency theory argues that this is an oversimplified approach and separates intrinsic from extrinsic motivation. Based on this theory the pay performance relationship might not be solely affected by a financial incentive and therefore the effects of higher flexible stimulation
18
might not be correlated with a stronger pay performance relationship. Acknowledging the two types of motivation the following hypothesis is formulated:Hypothesis 2: A higher percentage of flexible remuneration creates a disproportionate relationship between the intrinsic and extrinsic motivation of the agent. This leads towards lower organizational performance.
Concerning the agent’s time preferences in both agency and behavioral agency theory the value of money is discounted in the future. However, the way time is discounted differs.
Time is discounted exponentially within the agency theory and is done hyperbolically within the behavioral agency theory. The biggest difference can be explained by evaluating this theory in short term gains. The hyperbolic discount function makes the short term incentives even more preferable compared to exponential function. Within the topic of executive remuneration long term incentives are therefore less preferred by agents and the effect of them is less than argued for. In this study long term incentives are related to stock market incentives. Combined with the fact that stock market incentives are not only influenced by organizational performance we argue for a strong effect towards short term reward when stock market volatility is high.
Hypothesis 3: In years of high stock market volatility the pay performance relationship is weaker.
19
3. Method
Chapter two ended with the formulation of the hypotheses, this chapter builds further on how these hypotheses will be tested.
3.1 Underlying Case
The research focuses on CEO remuneration of top European companies and the correlation between organizational performance. As mentioned on in the research aim, this paper uses the behavioral agency theory to analyze contingencies that should explain the pay performance relationship. An alternative sample is chosen by selecting companies in Europe over the United States. Many important studies concerning this topic are focused on either the United States or Great Britain. This study does not attempt to find different results or aims to find abnormalities because of this sample. The ES50 index is used to select the companies listed in the sample. This index contains the 50 largest companies within the Eurozone. This index provides relevant companies from all over Europe to serve as a good reflection of large companies in Europe within all industries. Those 50 companies are based in twelve European countries: Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal and Spain.
3.2 Sample Description
The ES50 contains 50 companies, this is the index where the selected companies are selected from. Among those 50 companies, at the end of March 2016, 19 of them are French (which represent 37% of the values), 14 are German (32%), 6 are Spanish (10%), 5 are Italian (7%), 4 are Dutch (8%), 2 are Belgium (4%), and 1 is Finnish (2%).
Figure 3: Country reflection of the ES50 index
These companies are divided over 10 different sectors, Banks (14%), Industrial Goods &
Services (11%), Chemicals (9%), Insurance (8%), Personal & Household Goods (7%), Health Care (7%), Technology (7%), Oil & Gas (7%), Telecommunication (6%), and Food
20
& Beverage (6%). The sample includes 35 of the 50 values from the ES50. Subtracted are companies from the financial service sector within this paper.
Because of new regulations within the financial service sector banks and insurance companies are excluded. The European Union presented their amendment on the Basel 3 banking regulations affecting the banking and insurance sector (EU 2013). The European Union explains these reasons on the basis of three lessons that can be learned from the financial crisis. First of all the cooperation of monetary and fiscal and supervisory institutes around the globe should improve. Second of all there is a difference in the ability of companies to deal with shocks in the markets. This is mainly due to quality and the level of capital and its availability, liquidity management and the effectiveness of internal corporate governance. Third, by regulating banks across the borders a wider range of instruments could have been applied. The bail-out option was merely seen as the only option. When this was regulated across borders a wider set of measurements could be applied (EU 2013).
Regarding the remuneration and bonus payments in the financial sector these regulations put a cap on the bonus payments. The initial cap is a 1:1 ratio bonus salary payment versus fixed salary; this can be extended to a 2:1 ratio in certain conditions. When the shareholders vote in favor of this bonus payment extension the 2:1 ratio can be implemented. Six comments were made by Murphy (2013) as a response to the change in legislation of the European Union. The first one says that the fixed salaries increase and harms the adaptability of banks to market cycles. This will make the banks in the end more vulnerable.
Second, it does not lower the risk taking. Third, the cap on the bonuses will decrease the motivation to create value for the bank. Fourth, talent will flee outside of Europe. This will leave the European banking sector with less talented people. Fifth, the competitiveness of European banks compared to the outside world will decrease drastically. Sixth and final, the new measure does nothing with existing bonus contracts, therefore a one size fits all solution is not feasible for this problem. The decision concerning the exclusion of the financial service sector the first and third reason of Murphy (2013) mainly caused this decision.
21 3.3 Data Collection Procedure
The first step in the data collection procedure is collecting the organizational performance and executive remuneration data. The data concerning organizational performance was extracted from a database. Karlstad University has a subscription on a database named Amadeus from Bureau van Dijk. Amadeus contains comprehensive information of around 21 million companies across Europe. This database was used to extract the data of the independent variables for each performance measure. All of the variables of the four dimensional model were extracted from the database. Unfortunately not all data was available for every company in our sample. In two categories there was a significant amount of data lacking. This data was concerning the performance measures net sales and dividends paid. Additional data was gathered from a database called Statista, another online database where Karlstad University has a subscription on. For the data concerning the dividend per share the website Morningstar was used.
The data concerning CEO remuneration was not available in the databases and needed therefore to be collected with an alternative approach. This was done by extracting the data from the annual reports of the selected companies. First, the dependent variables listing information about CEO remuneration were divided into three different types of remuneration, separated in fixed salary, short term and long term remuneration. The reason to separate the salary components in just three categories and not in for example six categories as listed by Mallin (2013) are the differences in reporting formats by the companies in the ES50.
3.4 Research Design
The research conducted is a deductive analysis that is quantitatively performed. Mallin (2011) describes a deductive approach as a study that derives hypothesis from a theoretical construct consisting of previous research in the particular area. Consecutively, empirical data is tested to prove these hypotheses. Hypotheses 1, 2 and 3 were tested using a bivariate correlation analysis. The Relationships between scores (hypotheses 1, 2 and 3) were analyzed using the Pearson correlation model. The Pearson product-moment correlation coefficient, P, is a value between +1 and −1 inclusive, where 1 is total positive correlation, 0 is no correlation, and −1 is total negative correlation. When correlations are found and have a 0,05 or 0,01 significance level it is recognized as a relationship, this does not imply causation (Bryman & Bell 2013)
22 3.5 Methodology
Within the methodology the systematical analysis of the research is presented (Bryman &
Bell 2011). First the construct of the hypotheses are explained, followed by the measures of organizational performance. The chapter ends with the tests of the hypotheses.
3.5.1 The construct of the hypotheses
Based on the research design the hypotheses are tested as elaborated on below.
Hypothesis 1: In years of shareholder losses the link between shareholder remuneration and organizational performance is expected to be weaker in the year after the shareholders loss.
Shareholder losses years = Losses Shareholder Gains years = Gains Organizational Performance = Perf Absolute Value = AbsV
If AbsV[P(Losses,Perf)] < AbsV[P(Gains,Perf)] so H1 is true, the relationship is lower.
Hypothesis 2: A higher percentage of flexible remuneration creates a disproportionate relationship between the intrinsic and extrinsic motivation of the agent. This leads towards lower organizational performance.
Remuneration of Short term bonuses = Rshortterm Total remuneration = Rtotal
Organizational Performance = Perf Absolute Value = AbsV
If AbsV[P(Rshortterm,Perf)] > AbsV[P(Rtotal,Perf)] so H2 is true, the relationship is stronger.
Hypothesis 3: In years of high stock market volatility the pay performance relationship is weaker.
Remuneration of long term stock option = Rlongterm Total remuneration = Rtotal
Organizational Performance = Perf Absolute Value = AbsV
If AbsV[P(Rlongterm,Perf)] > AbsV[P(Rtotal,Perf)] so H3 is true, the relationship is stronger.
23 3.5.2 Measures of organizational performance
Profitability measures indicate how well a firm is performing in terms of its ability to generate profit. Hamann et al. (2013) found that the most significant values, that also take company size into account, are return per employee, return on sales and return on assets.
The return per employee looks at a company’s return in relation to the number of employees. In the database Amadeus the total number of employees and the revenues represented as net profit before taxes. This is calculated in the following way:
Return per employee: Net profit before taxes / Total number of employees
The return on sales is a ratio widely used to evaluate a company's operational efficiency.
Return on sale: Net profit before taxes / Sales
The return on assets (ROA) is an indicator of how profitable a company is relative to its total assets. ROA gives an idea as to how efficient management is at using its assets to generate earnings.
Return on assets: Net profit before taxes / total assets
Liquidity measures indicate the ability to pay operating expenses and other short-term, or current, liabilities. The liquidity ratios are a result of dividing cash and other liquid assets by the short term borrowings and current liabilities. They show the number of times the short term debt obligations are covered by the cash and liquid assets. The three measures used to measure the liquidity of our sample are Cash flow per employee, Cash flow on sales and cash flow return on assets. Cash flow per employee is a ratio which represent the amount of cash from operations per employee, measured as followed:
Cash flow per employee: Cash flow / total employees
Cash flow return on sales compares a company's operating cash flow to its net sales, which gives investors an idea of the company's ability to turn sales into cash. It is calculated as the cash flow on net sales.
Cash flow on sales: Cash flow / total sales
Cash flow return on assets measures how efficiently a business uses its assets to create a return or income. It is calculated as cash flow from operations on total assets. The higher the ratio, the more efficient the business is.
Cash flow return on assets: Cash flow / total assets
Growth measures indicate expansion and measure the growth of certain parameters compared to previous year(s). The employment growth also called the job growth figure is expressed as the gross number of jobs created in the company compared to the last year.
24
Employment growth: Employment year-Employment year/Employments year *100The sales growth is the amount by which the average sales volume of a company's products or services has grown compared to the last year.
Sales Growth: Sales year - sales year/sales year *100
The Asset Growth Rate of a company compares the total assets held at the end of the year compared to its Total Assets from the previous year.
Asset growth: Assets year- Assets year/ Assets year *100
Stock Market Performance measures the performance of a company and represents the measures closest to the interest of the investors. In contrast to the aforementioned profitability, liquidity and growth measures, the stock market performance focuses on future instead of the past. The measure representing stock market performance is the total shareholders return. This is measured as the change of stock price of a year compared to its previous year including the dividends received.
Shareholder return: Stock value end of current year- Stock value end of previous year + dividend current year
3.5.3 Test of the hypotheses
The first hypothesis focuses upon the years of shareholder losses the link between shareholder remuneration and organizational performance is expected to be weaker the year after the shareholder experienced the losses. After the data collection process two samples were created. One sample containing the pay/performance data from the years after a shareholder loss and the other one containing all the data within the sample. A bivariate correlation analysis was performed in SPSS to test those results.
With the second hypothesis an analysis of the relationship between fixed and flexible remuneration and the effects on organizational performance is made. First of all the distinction between fixed and flexible was calculated in the dataset stored in Excel. After the creation of this variable a bivariate correlation analysis was performed in Excel.
The third and last hypothesis focuses on stock market volatility. To calculate the stock market volatility the daily market data was extracted from Yahoo finance. The stock market volatility was calculated using the following formula:
The standard deviation refers to the annual standard deviation of the daily close values of the ES50 index. T is the factor time and relates in this case to the annual trading days, which are 252 days a year.
25
4. Empirical findings and results
In order to create a better understanding of key variables used in the hypotheses and improve the readability, the descriptive statistics are briefly explained. As pointed out by Podsakoff and Dalton (1987), researchers do not frequently post stability measures of their results. By doing so it also allows the reader to create his or her own critical reflection.
Three important variables are briefly highlighted before the presentation of the results.
These variables are: total salary, flexible remuneration and the fixed flexible. The range of the total salary varies from 1.027.355 to 16.163.400. This range gives an indication of the wide variety of the CEO remuneration within the ES50. However most of the results are concentrated around the mean of 5.191.203. With a standard deviation of 2.890.044 the data overall sample seems to be in balance, with a couple outliers. By taking a further look at the flexible part of the remuneration also a rather outstretched sample can be recognized with a minimum of 284.026 and a maximum of 13.863.400. The flexible part of the remuneration seems to be centered on 58 to 78 percent of the total remuneration. The results per hypothesis are listed below. All of the results that are shown are listed and calculated in Euro (€).
Hypothesis 1
In years of shareholder losses the link between shareholder remuneration and organizational performance is expected to be weaker the year after.
The data of the first hypothesis was examined in two different ways. First the data was separated in Shareholder gain- and shareholder loss positions referring to figure 1 in the literature review. The first analysis focused on the differences between shareholder remuneration and organizational performance during the years of shareholders losses and during the years of shareholders gains. In figure 2the results of the SPSS bivariate correlation analysis is shown.
26
Figure 4: Correlation between organizational performance factors and two indicators of the remuneration (variation of the salary compare to previous year and the total salary) separately during the years of shareholders losses and years of shareholders gains between 2010 and 2013.
As can be seen in figure 2 the overall results show no promising new insights nor a structural pattern. There is neither correlation during the years of shareholders losses nor during years of shareholders gains. However, there is one significant relationship that can be found. The relationship between return per employee and the total salary during the years of shareholders losses was found to be a significant negative relationship (-0,407).
This correlation is significant at the 0.05 level (2-tailed). This result suggests that for a weaker return per employee, the total salary of the CEO is higher.
The second comparison analyses the correlation between shareholder remuneration and organizational performance during the years of shareholders losses and to our complete sample. In figure 3 the results of the SPSS bivariate correlation analysis is shown.
27
Figure 5: Correlation between organizational performance factors and two indicators of the remuneration (variation of the salary compare to previous year and the total salary) separately during the years of shareholders losses and to our entire sample between 2010 and 2013.
The data concerning shareholder losses is identical to the data used in the first comparison.
And so the relationship between return per employee and the total salary during the years of shareholders losses was found to be a significant negative relationship (-0,407).
Concerning the second part of the figure the Sig. 2‐tailed level indicates four times .000 which shows that there is significance between the total salary of the CEO and organizational performance factors. In details, there is significance between the total salary of the CEO and return on sales and the relationship is a positive 40,8%, which means that as one variable goes up or down so will the other one. There is also significance between the total salary of the CEO and return on assets and the relationship is a positive 40,9%.
Then we found the significance between the total salary of the CEO and the cash flow return on assets and the relationship is a positive 36,1%. Finally, there is significance between the total salary of the CEO and shareholder return and the relationship is a positive 36,7%.
28
Hypothesis 2A higher percentage of flexible remuneration creates a disproportionate relationship between the intrinsic and extrinsic motivation of the agent. This leads towards lower organizational performance.
Figure 6: Correlation between organizational performance factors and the part of flexible remuneration.
As can be seen in figure 4 the overall results there is not a pattern recognizable pattern within higher flexible remuneration and the linkage between organizational performance.
However, there is one significant relationship that can be found. The relationship between return on sales and the flexible/fixed salary ratio was found to be a significant relationship (0,238). This correlation is significant at the 0.05 level (2-tailed). This result shows that here is a relation between flexible part of the CEO remuneration and the return on sales.
Hypothesis 3
In years of high stock market volatility the pay performance relationship is weaker.
The annual volatility as shown in figure 5 displays rather small differences between the annual volatility between the years 2012, 2013 and 2014. According to these volatility numbers it is expected to find the lowest correlation in 2012 and the highest correlation in the year 2013
Year Volatility
2012 10,75
2013 8,12
2014 9,02
Figure 7: Annual volatility.
The variables related to the CEO remuneration can be divided into an interval value reflected by the total salary, and a ratio reflected by the change of total salary.
29
Change of total salaryIn 2012, the relationship between the change of salary compared to previous year and return on assets was found to be a significant relationship (0,343). Another significant relationship between an organizational performance factor and total salary change is the factor of cash flow return on assets (0.391). Both correlations are significant at the 0.05 level (2- tailed). In 2013, the relationship between the changes of total salary compared to previous years significantly negative relations were found with all of the organizational growth factors. Employment growth was found to be a significantly negative relationship (-0,431), also sales growth was found to be a significantly negative relationship (-0,513) and the same finally for assets growth (-0,513). All those correlations are significant at the 0.01 level (2-tailed). In 2014 shareholder return was found to be a significant relationship (0,349), which means that if the salary increases compared to the previous year the shareholder return also increases. This correlation is significant at the 0.05 level (2-tailed).
Also, the relationship between the fixed on flexible ratio and return on sales was found to be a significant relationship (0,355).
Total salary
Comparing the organizational performance indicators with the static salary gives us an entire different perspective than relating organizational performance to salary change. In 2012 significant relations between total salary and the organizational performance factors were found for return on sales (0,464), return on assets (0,411), employment growth (0.417), Sales growth (0.381), asset growth (0.390) and shareholder return (0.505). In 2013 no significant relationships were found related to organizational performance. In 2014 significant relations between total salary and the organizational performance factors were found for return on Sales: (0,498), return on assets: (0,594), cashflow return on assets:
(0.571), asset growth: (0.381) and shareholder return: (0.367)
30
5. Discussion and analysis
5.1 Theoretical discussion and analysis
In the theoretical discussion the results are analyzed and brought into relation with the theory of the theoretical framework. The hypotheses are discussed upon in ascending order.
Hypothesis 1
The first hypothesis tests whether in years of shareholder losses the link between shareholder remuneration and organizational performance is weaker in the year after a shareholder loss. The organizational performance was tested with two salary variables, salary growth and total salary. The measure that is more supportive in explaining this hypothesis is the total salary. The results that were found by comparing the years of shareholder losses with salary growth were significantly weaker than the years of shareholder losses. The results indicate that there might be a causation between loss averse behavior and the decision making by investors that influences the pay performance relationship. This insinuates that the preferences of the principal, as argued for in the agency theory as being neutral, might not be so neutral after all (Krause et al. 2014). In the behavioral agency theory constructed by Pepper & Gore (2015) the loss averse behavior of the principals is argued for. However, in their propositions a rather inductive approach is drawn in contrast to the statistical analysis where a rather general approach is applied.
Therefore it is difficult to draw a hard conclusion from these results. The findings of Krause et al. (2014), based on an experiment are supportive for this theory. By taking into account personal behavioral factors and analyze it from a general perspective, a significant relationship is found. However, something that should be acknowledged is that the moment of obtaining the shares is not taken into account in this thesis. Regardless of the positive outcome of the first hypothesis, stronger results might be found when moment of obtaining a stock is calculated in. On the other hand, results are usually annually reported on and for example institutional investors might rather be reviewed upon their quarterly/annual results rather than their moment of entry. The results of this analysis should be seen as preliminary findings directed towards loss averse behavior. Additional research is needed to confirm this over time.
Hypothesis 2
The second hypothesis tests if a higher percentage of flexible remuneration creates a disproportionate relationship between the intrinsic and extrinsic motivation of the agent
31
which leads towards a weaker pay performance relationship. As can be extracted from the results, there is only one significantly positive relationship that rather seems incidental than explaining the construct. The overall results of the second hypothesis show no clear results that support that a higher flexible remuneration has a stronger relationship with organizational performance. This is an indication that more flexible remuneration does not necessarily accelerates organizational performance. By testing this hypothesis a general approach to the topic is examined. This means that purely intrinsic and extrinsic motivation cannot solely looked upon as the explanation. A possible further direction, which is argued by Pepper & Gore (2015), suggests a personal approach where the intrinsic and extrinsic motivation should be evaluated on pay effort curves from individuals. However, this would lead away from a general theory and deductive approaches would be more difficult in explaining certain behavior. Although, when the variance of the reference levels of the motivation of executives would are established boundaries can be created to measure upon.Besides giving an indication about motivation and interest alignment, the results of this hypothesis also support the claims made by Frey and Jegen (2001) and Sliwka (2007) for a possible broken pay setting process. Therefore we argue that specific linkages should be sought in specific parts of the executive remuneration, as for example the long term/short term relationship or cash versus stocks. Because of the general approach this study is limited in seeking the explanation by just comparing the fixed flexible relationship. By expanding the understanding of the impact of motivation of organizational performance, the relevancy of the question about the relationship between internal and external motivation and how this can be explained within the remuneration construct is simple. If the agency construct can be explained by behavioral factors an optimal salary range can be created where executive remuneration can be tested upon. A possible other explanation could be that non-organizational performance factors determining executive. This could be another interesting view upon executive remuneration. Since non-financial performance measures became more interesting after the publication of the balanced scorecard by Kaplan and Norton (1995).
Hypothesis 3
The third hypothesis tests if in the years of high stock market volatility, the pay performance relationship is weaker. After calculating an annual volatility measure for the years 2012, 2013 and 2014, a comparative analysis was made between these years. In the year of 2013, when stock market volatility was at the lowest point, the weakest pay performance relationship was found. This result contradicts the initial assumption and
32
therewith our hypothesis. From a behavioral economic perspective, the assumption was made that in years of high stock market volatility the focus of the CEO would even have a stronger hyperbolic effect on time preference and therefore weaken the pay performance relationship, according to the effect found by Ainslie (1991). Hamann et al. (2013) argued for avoiding years of high stock market volatility because it would influence the pay performance relationship. In the years that are analyzed, relatively small differences among the volatility are recognized, which could be a reason for the contradicting results. In the years 2012 and 2014 the pay performance relationship is found stronger than the original sample, which clearly indicates that the results of 2013 influence the sample. However, the reason of this influence cannot be explained by stock market volatility. Despite of the results from Hamann et al. (2013), who were not coming from only one particular industry, a possible explanation of a stronger pay performance relationship may differ within certain industries. As for example the financial services sector, which is excluded from the sample used in this study, who is also result wise more related to stock market performance. In this test the effect of stock market performance was not isolated and for example solely tested in relations to (long term) stock option and share plans. Therefore, other factors might play an important role in explaining these variations. This does however encourage to zoom in to the year 2013 to find a better understanding of these results and develop a better understanding of these movements.The topic of executive remuneration and the pay performance relationship has been a widely discussed subject. As one of the first studies of connecting the behavioral agency theory in explaining the pay performance relationship this study contributed to the executive remuneration construct. By implementing behavioral theory in order to explain the pay performance relationship this study provided a new foundation for scholars to build further on. Rather than altering existing theory a new foundations for theorizing about executive remuneration is provided. This study suggest that motivation and loss aversion play a role in explaining the pay performance relationship. The evidence regarding the effects of stock market volatility does not to hold in our findings. The findings provide an explanation for the inconsistent results in previous research, reflected upon in the large meta studies (Tosi et al. 2000; van Essen et al. 2015). An implication that should be considered by reading this study is the sample that is coming from a different geographical location. Hence, the comparability might be affected as elaborated on the previous chapter.
Therefore future studies need to place this theory in a different context to test if a general understanding can be created.
