Ever since the financial and economic crisis, risk-taking and CEO compensation have been a popular topic. Managers frequently must select from a large amount of projects or investments that vary greatly in both risk and expected return. For example, managers frequently engage in capital budgeting decisions in which they have to evaluate and select investments that differ in the timing, magnitude, and riskiness of cash flows (Bonner, 2002). In these situations, the accounting performance measurement and reward system not only needs to motivate high levels of effort from employees, but also needs to encourage the appropriate level of risk-taking.
According to a survey by PriceWaterhouseCoopers (2008) of financial services professionals the three most important factors to create the financial crisis in 2008 were excessive risk taking, mispricing of risk and rewards systems. Also, former Federal Reserve Vice Chairman Alan Blinder blames the financial crisis on the incentives built into compensation plans of many financial firms have encouraged excessive risk-taking (Larcker, 2014). Therefore, it is necessary to examine which incentive schemes induce managers to take appropriate levels of risk while concurrently motivating high levels of effort (Sprinkle, 2003).
Following agency theory (Jensen & Meckling, 1976), agents are assumed to be strictly rational and risk-averse. Therefore the theory expects agents not to react on performance based compensation or other incentives in terms of risk taking. If a principal wants its agent to be more risk-taking, not only an incentive regarding performance or effort, but also an incentive related risk must be provided, such as a risk-premium (Holmstrom, 1989).
Contradictory to agency theory, scholars from the field of behavioral science argue that risk does increase when performance based compensation is implemented. Reason for this is that an individual will set a goal in order to receive the desired compensation and according to the findings of Larrick (2009) individuals are significantly more risk-seeking when they set a specific, challenging goal. Frank (1935) describes these subjectively established goals as the aspiration level of an individual. So, performance based compensation motivates individuals to do better, not only their effort, and thus performance, will be bigger, but they also increase their aspiration level and are therefore willing to take bigger and maybe even excessive amount of risk.
Many companies base their incentive scheme and performance evaluation systems not only on the evaluation of performance of the employees’ themselves, but on their performance relative to the performance of their peers. Relative performance evaluation (RPE) is an evaluation method in which an individual’s performance is compared to the performance in a group of the individual’s peers (Financial Times Lexicon 2016).
Festingers’ social comparison theory (1954) emphasizes that relative performance evaluation can cause individuals to be more competitive and thus be more risk-seeking. According to this theory individuals have a desire for self-evaluation and to know what they are capable to do. It also states that individuals have the drive to compare themselves to others in order to evaluate their abilities. Because individuals are not able to objectively test their decisions and abilities, they feel the need to use performance of others to compare themselves to. Performing better than others leads to positive feelings such as pride, while performing worse than others leads to negative feelings such as shame (Hannan, 2013). In order to maintain a positive self-image, individuals are motivated to perform better than their peers and thus will put more effort into their work. However, it is also likely that because of this urge to perform better, individuals are also willing to take excessive amounts of risks in order to do better than their peers.
In this paper, I propose that relative performance evaluation (RPE) and performance based compensation are factors that influence the amount of risk an individual is willing to take mediated by the level of aspiration. I also propose that as the two theories contradict each other, this paper will make clear which of these theories is leading. The research question for this thesis will be: “What is the effect of Relative Performance Evaluation on investment risk taking, effort and performance of employees?”
In order to answer this research question I would like to conduct a 2 x 2 between-subjects factorial design. The data will be collected using an experimental method. The first manipulation is RPE, two of the experimental groups receive RPE and the other two receive only individual performance evaluation. The second manipulation is compensation type, two of the experimental groups will be compensated according to their performance and the other two experimental groups will receive a flat rate compensation.
Chapter 2 – Literature review and hypothesis development
In this chapter the relationships between the variables of this research will be explained. First of all, the anticipated relationship between performance based compensation and risk-taking will be discussed. Second, attention is paid to the hypothesized effect of relative performance evaluation on risk-taking. Third, the expected interaction effect of the two independent variables on the dependent variable is described. Finally, the conceptual framework with the relationships between the discussed concepts are outlined in an graphical overview.
Effect of compensation based pay on risk taking
As discussed in the introduction, risk-taking of managers is a popular topic since the economic crisis. Taking the appropriate amount risk is key when individuals have to select a project or investment. In these situations, the accounting performance measurement and reward system not only needs to motivate high levels of effort from employees, but also needs to encourage the appropriate level of risk-taking.
Performance based compensation can be described as a system that relates rewards of an employee to the performance of the organization or individual. Performance based pay is used to motivate employees to put more effort in their work and to equalize the individual goals of the employees with the goals of the organisation (Economist, 2009). Theories about the relationship between the independent variable compensation based pay and the dependent variable risk-taking are contradicting. The first theory to discuss is the agency theory.
Agency theory can be defined as a contract under which one person (the principal) engages another person (the agent) to perform a service on their behalf (Jensen & Meckling, 1976). This contract requires that some decision-making authority is delegated to the agent and it focuses on conflicts of interest between managers or employees, being agents, and owners, being principals. According to the agency theory one of the main reasons that managers or employees tend to be risk-averse is that increased risk can potentially affect a person’s personal success rate. The theory is based on the assumption that agents are (1) risk averse, (2) self-centered and that (3) their interests differ from those of the principal. Agency theory is part of conventional economic theory and it assumes agents to act in their self-interest, maximize their own personal interests and minimize the possibility of personal loss (Jensen & Meckling, 1976). The challenge for the principle is to induce the agent to act in the interest of the principles and thus to act in order to increase firm value and firm performance. The discrepancy of interests leads to agency conflicts. This is why the agency theory is often referred to as the agency problem.
To solve the agency problem, a possible solution is to monitor the actions of the agents. However, a better solution is to make sure that the agents best interest is equal to the best interest of the principal by providing the correct incentives and compensation. According to Aggarwal and Samwick (1999), the best way to do this is to tie individual compensation to performance. Conventional economic theory posits that performance evaluation should only have an effect on effort when compensation is based on performance (Hannan, 2013; Kandel and Lazear, 1992).
A fundamental assumption of agency theory is that individuals are fully rational and that when their task does not contribute to their own economic wellbeing, they will not feel aspired to exert effort on a task (Bonner, 2002). Rational individuals can be defined as people that always choose the option that makes them better off (Jensen, 1994). When an individual is rational, he or she is expected to be risk-averse because This means that agents are assumed to act in their self-interest and thus not in the interest of their superiors or peers interest unless the agents’ personal wealth is maximized. Because agency theory assumes agents to act in their self-interest, it is expected that when performance based compensation is applied, agents will put more effort into their performance compared to when flat wages are applied. However, because performance based compensation the incentives is not directly linked to risk-taking, it is expected that risk-taking is not affected by the compensation type. In short, agency theory expects agents not to react on compensation or other incentives in terms of risk-taking. If a principal wants its agent to be more risk-taking, not only an incentive regarding performance or effort, but also an incentive related risk must be provided, such as a risk-premium (Holmstrom, 1989).
However, a contradicting conclusion can be drawn when looking at goal-setting theories. When a performance based compensation is provided, an individual will set a goal in order to receive the desired compensation. For instance, a sales employee will be more motivated and thus will set a higher goal to sell products when he receives compensation per item sold compared to when he receives a fixed monthly salary. According to the findings of Larrick (2009), individuals are significantly more risk-seeking when they set a specific, challenging goal. Goal-setting motivates individuals to achieve the goal and in order to do so, individuals are willing to take more risks. Frank (1935) describes these subjectively established goals as the aspiration level of an individual. If the individual outperforms the aspiration level, this will be seen as a success and when performance does not meet the aspiration level this will be seen as a failure. In other words, performance based compensation motivates individuals to do better, not only their effort, and thus performance, will be bigger, but they also increase their aspiration level and are therefore willing to take bigger and maybe even excessive amount of risk.
In this paper, the relationship between performance based compensation and risk-taking will be hypothesized according to the goal setting. This leads to the following hypothesis:
H1: Performance based compensation leads to more risk-taking because of a higher aspiration level
Effect of relative performance evaluation on risk-taking
Many companies base their incentive scheme and performance evaluation systems not only on the evaluation of performance of the employees’ themselves, but on their performance relative to the performance of their peers. Relative performance evaluation (RPE) is an evaluation method in which an individual’s performance is compared to the performance in a group of the individual’s peers. This method ensures that the individual will not be held responsible for factors beyond its control such as natural catastrophes, general economic conditions, stock market volatility and government regulation (Financial Times Lexicon 2016).
The social comparison theory, which is part of the behavioral field of science, argues that, contradictory to the conventional economic theory, monetary incentives are not the only motivator for individuals to feel aspired to put effort into their work. It states that intrinsic motivation also plays a big role. The behavioural theory behind relative performance evaluation is the social comparison theory (Festinger, 1954). This theory states that because individuals are compared to each other, their motivation regarding effort and performance increases. This is called the motivation effect. The social comparison theory emphasizes that individuals have a desire for self-evaluation and to know what they are capable to do. According to the social comparison theory individuals have the drive to compare themselves to others to evaluate their abilities. Also Heslin (2003) argues that individuals use others as referents to verify whether their performance is successful. Social comparison does not only play a role in individuals personal life, but also in the organizational process of evaluating performances of employees. According to Mumford (1983) supervisors use social comparison to evaluate the performance of their subordinates relative to each other.
Because individuals are not able to objectively test their decisions and abilities, they feel the need to use performance of others to compare themselves to. Performing better than others leads to positive feelings such as pride, while performing worse than others leads to negative feelings such as shame (Hannan, 2013). In order to maintain a positive self-image, individuals are motivated to perform better than their peers and thus will put more effort in their work.
As discussed earlier in this chapter, individuals’ aspiration regarding effort and performance increases when their performance is compared to that of their peers. This is why firms commonly use RPE even when it is not linked to their compensation (Hannan, 2013). For instance, sometimes sales companies inform their sales men about the realised revenue or sales margin per employee in the form of a ranking list in the canteen because this stimulates the competitiveness among the employees. In line with the social comparison theory, the second hypothesis can be stated as follows:
H2: Relative performance evaluation leads to more risk-taking because of a higher aspiration level
Interaction effect
Because both independent variables have an effect on risk-taking through the mediating variable aspiration level, I expect an interaction effect. According to Frederickson (1992), compensation based on RPE causes individuals to be even more competitive and thus have an even higher aspiration level compared to individuals who only receive RPE. This expectation leads to the following hypothesis:
H3: The difference in risk-taking between situations in which RPE is provided and situations in which individual performance is provided is larger when compensation based on performance is provided compared to when compensation based on flat rate compensation is provided.
However, there is possible limitation for this hypothesis. When RPE and performance based compensation are combined, it is possible that there will only be a small interaction effect or possibly even no interaction effect because the maximal level of aspiration is already reached.
Conceptual framework
Figure 1: Conceptual framework
The figure above shows the conceptual framework of the research design. All the theoretic concepts and their relationship are shown in a clear overview. It also, the three hypotheses discussed in the previous paragraphs and their relationship are shown clearly.
Chapter 3 – Research design and data
In this chapter the research design of thesis is discussed. First of all the research methodology and the procedure to collect data is discussed. Secondly, the statistical tests to test the hypotheses mentioned in chapter 2 of this proposal. Finally, attention is paid to the construct, validity and reliability of the methodology.
Manipulation
For this research I will conduct an experiment with a 2 x 2 between-subjects factorial design. The sample of the population of this experiment consists of students. In total, approximately 80 individuals will participate in the experiment. According to the two manipulations the participants will be assigned to four different experimental groups. The first manipulation is evaluation type, two of the experimental groups receive RPE (group 1 and 2) and the other two will receive individual performance evaluation (group 3 and 4). The second manipulation is compensation type, two of the experimental groups will be compensated according to their performance (group 1 and 3) and the other two experimental groups will receive a flat rate compensation (group 2 and 4). For the evaluation type variable, it is key to create a situation in which the participants feel like they are compared and competitive to each other. Therefore I have chosen to perform the experiment in pairs of two in order to implement a realistic relative evaluation type. In order to simulate a working-place environment in which colleagues “compete” with each other, I will only select pairs of participants who know each other and are not strangers. According to several studies individuals are more competitive to people they know than to strangers. For instance, Zuckermann & Jost (2001) state that individuals feel more threatened by the success of friends than that of strangers. Also, individuals provide less help to friends than they do to strangers in competitive tasks that are self-relevant (Tesser, 1988). The experimental groups with the RPE manipulation will be informed before the start of the experiment that they will be evaluated relative to their peer. Also, the participants will be able to see each other so the competition-effect will be stronger. However, the participants will conduct the experiment separately so the pairs of participants are not allowed to discuss the questions with each other.
The experiment will start with a short instruction. The participants will be told how they will be evaluated and compensated depending on the experimental groups that they are assigned to. Also, they will be instructed to imagine that they work at a trading company and that they are the manager of the financial department. I choose this position because this is a position wherein an employee is likely to be exposed to investment decisions. The approach to evaluate and compensate the different experimental groups is discussed below:
– Group 1 – RPE & Compensation based on performance: The participants will be evaluated relative to each other, so one of them will be the “winner”. The winner will receive €2.
– Group 2 – RPE & Flat rate compensation: The participants will be evaluated relative to each other, so one of them will be the “winner”. Both participants will receive €1 for their participation.
– Group 3 – No RPE & Compensation based on performance: The participants will be evaluated individually for their score, there will be no “winner”. If the participants have performed “well” they will receive €2. Performing “well” means that the participant must solve at least two out of three cases correctly. The participants will be informed that approximately 50% of the previous participants have reached this goal.
– Group 4 – No RPE & Flat rate compensation: Each participant receives €1 for their participation, there will be no winner. Scores will be provided to the participants individually.
Experimental task
The experimental task of the experiment will be timed for each of the participants separately by the researcher. The participants will be informed that they can take all the time they need to perform the experimental task. The experimental tasks consists of three case questions. In the first part of the case the participant is asked to decide between four investment options, varying from a very risky decision to a very low risk decision. For the second part of the case, the participant is asked to answer some questions regarding investing that require calculation, such as calculating the investment return. The participants will be informed that they will be evaluated on performance for the case but in reality the evaluation will only be based on the second part of the cases . The answers for the first part of the case regarding the investment decisions will only be used to determine the risk-taking of the participant. To measure the effort the participants put into the experimental task, the reaction times of the participants will be timed.
Even though Kahneman and Tversky (1973) have shown that problems that require calculations can influence the results of the study more compared to for instance puzzles or spelling problems, I choose to let the participants solve questions that require calculations. The reason for this is that I want the participants to feel like they are evaluated for the entire case, including the investment decision, while actually their performance evaluation depends on the investment calculations. The best way to mask this, is to combine the investment calculations and risk-taking decision in one case. This also makes it possible to let the participants imagine that they are employees or managers of a company, which improves the generizability.
The participants will also be asked to answer a couple 5-point likert scale questions in order to check for unexpected results. The participants will also answer a couple of 5-point likert scale questions to test their aspiration level. According to the regulatory focus theory (Higgins, 1997) the way in which an individual attempts to achieve their goal depends on the regulatory orientation of the individual. Regulatory orientation can be defined as the way that an individual chases a goal in a way that the personal values and beliefs are sustained (Avnet, 2003). The theory states that individuals can have two different types of regulatory orientation, namely a promotion-focus orientation or a prevention-focus orientation. To test whether or not the regulatory focus of an individual has an effect on the aspiration level of the participant, some questions to test their regulatory focus will be asked. Another check for an unexpected result will be a check on the risk-attitude of the participant. If the risk-attitude of the participant, it is possible to check whether this has an effect on the results of this study.
The last part of the experiment consists of a couple of demographical questions such as gender and age and some manipulation check questions such as whether or not the participants believe that they were evaluated relative to their peer and whether or not they believe their compensation is based on their performance. Second, the participants will also be asked about the amount of effort that they have put into the experiment. The results of this question can be used to confirm whether the reaction times of the participants are consequent with the opinion of the participant about their effort. Finally, the experiment ends with some factual questions to verify whether the survey was understood correctly.
Statistical tests
To analyse the data retrieved by performing this experiment, several statistical tests will have to be performed. The software I will use for these analyses is SPSS. First of all, descriptive statistics will be provided. Also, some general checks such as a manipulation check will be performed using the Cronbach’s alpha of the two manipulation check questions. Also a normal distribution check will be performed to confirm that the four experimental groups are normally distributed.
The first hypothesis predicts a positive relation between the independent variable compensation type and the dependent variable risk-taking through the mediating variable aspiration level aspiration level will be tested using an ANCOVA test. For testing the second hypothesis, the relationship between evaluation type and risk-taking through aspiration level, also an ANCOVA test will be used. The direct effects of the independent variables on risk-taking will be tested using two separate one-way ANOVA tests. Another option to test hypothesis one and two is to perform a mediation analysis. To test for a interacting effect a two-way ANOVA test will be used. To test the entire model, a regression analysis will be performed.
Construct validity
Construct validity has to do with the degree to which the measurement of a concept in a study is a correct reflection of the theoretical constructs mentioned in the study. The difference between construct validity and external validity is that construct validity involves generalizing from the measures in the study to the concept or theory of the measures while external validity focuses on generalizing from the context of the study to other individuals. In other words, construct validity translates a construct or theory into an operationalization (Social research methods, 2006). The construct validity of the dependent variable, risk-taking, depends on the five decisions regarding risk-taking the participants will have to take. The other ten questions about business problems are part of the experiment for another reason, which is comparing the performance of the participants and masking the risk-taking problems. To measure the construct “investment risk-taking”, questions that are comparable with real-life business investment decisions. An example is: Below four different investments with their corresponding expected returns are described. Your supervisor asks you to make a decision, which one do you think is the best decision?
Internal and external validity
Internal validity is the degree of causality that can be proven for a relationship between the independent and dependent variable (Neumann, 2011). The internal validity can be threatened by several factors. Confounding is minimized by only manipulating the two independent variables and keeping everything else the same. Also, selection bias is avoided by assigning the participants to the experimental groups randomly. Because a post-test design is used, participants were not informed about the purpose of the study and therefore also testing bias is reduced. Instrumentation biases are also excluded from this study because I will use a pre-conceived plan and script.
External validity focuses on the question whether or not the results of the study are generizable to other empirical settings (Bryman, 2007). Because a big variety of students will participate in the experiment, external validity of this study is enlarged. This will be checked by testing the distribution of participants across experimental settings. Moreover, the results will be generizable to employees in general, but also to managers because, in general, the students are the managers of the future. A weakness for the external validity is that participants will not be engaged by random sampling.
Reliability is mainly concerned with issues concerning the consistency of the measures and with the question whether or not the results of the study are repeatable. The reliability of this study can be checked using the Cronbach’s alpha.