UNIVERSITEIT VAN AMSTERDAM
MASTER’S THESIS
Final version
The influence of tenure, perceived wealth-based
inequity and perceived personal wealth on
gaming incentive systems
Executive Programme in Management
Studies – Strategy Track
H. Hylkema
(10730672)
June 17, 2016
STATEMENT OF ORIGINALITY
This document is written by Student Hanke Hylkema 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.
CONTENT Abstract ... 4 Introduction ... 5 Literature review ... 7 Agency theory ... 7
Incentive systems and gaming ... 8
Tenure ... 10
Unfairness – wealth-based inequity ... 11
Personal wealth ... 12 Visualization ... 14 Methodology ... 14 Research design ... 14 Level of analysis ... 15 Sample ... 15 Research procedure ... 16 Variables ... 19 Dependent variables ... 19 Independent variables ... 20 Data analysis ... 23 Results ... 26
Conclusion and discussion ... 31
Managerial implications ... 38 References ... 41 Appendices ... 45 Appendix I ... 45 Appendix II ... 49 Appendix III ... 50 Appendix IV ... 51
ABSTRACT
Gaming incentive systems is about employees exploiting incentive systems for their own personal benefit. This negative effect of incentive systems got growing attention in the literature and is argued to outweigh the positive effects of incentive systems. This research enriches the literature by looking more into the factors that lead employees to gaming incentive systems. It is argued that tenure, perceived wealth-based inequity and perceived personal wealth are factors in driving employees into gaming incentive systems. Tenure is argued to be a driver, because of the experience employees’ gain when working with the incentive system. Significant evidence is found supporting this hypothesis. Perceived wealth-based inequity is shown to be a driver as well, because of the feeling of unfairness it brings. Overconfidence is proven to have a positive moderating effect on the relation between perceived wealth-based inequity and gaming. Finally, perceived personal wealth is argued to be a factor as well. Significant evidence is found that perceived low personal wealth drives people to gaming. The findings raise questions about the nature of incentive systems and the assumptions that they are built on. Incentive systems are designed to align the goals of the employee and the firm, but could also bring out behavior that would not have existed without an incentive system. The results and conclusions of this research raise the question whether incentive systems are built on the wrong assumptions and actually lead people to gaming while they would not have gamed under the right conditions.
INTRODUCTION
The power of incentive systems and the effect on employee behavior and firm performance has been excessively discussed in the literature (Baker et al., 1987; Holmstrom, 1979; Levinthal, 1988). It is argued many times that employees “game” incentive systems in numerous ways for their personal benefit. Research shows that this harms the organization, because the employee pursues his individual goals instead of the firm’s goals (Kreps, 1997; Larkin et al., 2012; Ordonez et al., 2009). Although there have been many studies on the positive and negative effects of incentive systems and the ways in which employees can game incentive systems, there is little research on the factors that lead to gaming incentive systems.
Agency theory assumes that people are self-interested and will therefore pursue their individual needs first (Eisenhardt, 1989). This assumption is the basis to the principle-agent problem, which states that the goals and risk preferences of the principle and the agent do not align. Incentive systems play a key role in overcoming these differences and in aligning the interests of the principle and the agent. The positive effects of incentive systems, which include motivating employees (Holmstrom, 1979; Levinthal, 1988), enhancing productivity (Ichniowsky, Shaw, and Prennushi, 1997; Obloj & Sengul, 2012), and attracting high-ability employees (Clinch, 1991; Levinthal, 1988), are offset by negative effects of incentive systems that in fact harm the organization (Kreps, 1997; Larkin et al., 2012; Ordonez et al., 2009). The most important negative effect of incentive systems is the exploiting or “gaming” of incentive systems.
Gaming incentive systems means that employees learn how to exploit the incentive system and how to use it for their personal benefit (Obloj & Sengul, 2012). When this happens, the main objective of the incentive system - to align the interests of the organization and the
employee - is not met. Employees can game incentive systems by gaming the realization of targets or by gaming the setting of targets (Jensen 2003). The literature identified diversion of effort to activities that are rewarded (Larkin, 2014; Oyer, 1998; Pierce, 2012) and exploitation of the rules to improve apparent performance (Gravelle, Sutton, and Ma, 2010) and timing (Larkin, 2014; Oyer, 1998) as ways to game incentive systems. Much research is performed on these topics, but the factors that make employees game incentive systems remains underexplored.
This research studies three possible factors that make employees game incentive systems: tenure, wealth-based inequity and personal wealth. Recently, Frank and Obloj (2014) found a relation between firm-specific human capital and gaming. These findings also indicate a possible relation between tenure and gaming. Larkin (2014) found that tenure has a positive relation with gaming in the form of timing. In this research, these findings will be broadened by focusing on gaming of targets and gaming of realization but not in one specific form. The second factor examined in this research is wealth-based inequity. Building further on the findings on the relation between wealth-based inequity and unethical behavior (Gino & Pierce, 2009a, 2009b, 2010; John et al., 2014), this study aims to prove the relation between wealth-based inequity and gaming. Following agency theory and logic, personal wealth is the last factor examined in this research.
Although there has been much research on different forms of incentive gaming, there is not a lot of research on the factors leading to gaming incentive systems. This paper aims to study how tenure, wealth-based inequity and personal wealth are factors in driving employees into gaming incentive systems by answering the following research question: How do tenure, wealth-based inequity and personal wealth influence gaming incentive systems?
In an attempt to answer the research question an experiment will be conducted. Gaming incentive systems is a sensitive topic, because it can be seen as employees behaving
unethically. This can lead to people not responding honestly when asking about their gaming behavior. An experiment makes it possible to create specific circumstances in order to find out more about gaming behavior. This is expected to lead to honest responses, because it is in an experimental setting.
THEORETICAL FRAMEWORK
In order to answer the stated research question several factors have to be examined. First of all one should understand why incentive systems are needed in the first place and what the underlying assumptions are. Agency theory (Eisenhardt, 1989; Jensen & Meckling, 1976; Ross, 1973) will provide these insights. Secondly, the recent literature on incentive systems and gaming will be examined in order to get a clear understanding of the current state of research. Finally, as this paper is about specific drivers of gaming incentive systems, these drivers are examined in order to explain why these specific drivers are chosen.
Agency theory
Risk sharing literature (Arrow, 1971; Wilson, 1968) describes risk sharing among individuals and groups, which occurs when cooperating parties have different attitudes towards risk. Agency theory follows up on this risk-sharing problem by describing the relationship between a principle and an agent in which the principle has the power to delegate authority to the agent (Eisenhardt, 1989). Because of the delegation, the agent affects the principle’s welfare. When the goals of the principle and the agent do not align, the agency problem occurs (Jensen & Meckling, 1976; Ross, 1973).
According to Eisenhardt (1989), agency theory is about solving two problems that can occur in a principle-agent relationship. The first problem occurs when goals of the principle and agent do not align and when it is hard to know how the agent is behaving. The main
assumption of agency theory is that people are self-interested (Eisenhardt, 1989). Therefore, when goals are conflicting and the principle does not know exactly what the agent is doing, there is room for opportunism. This is why it is important to align principle-agent goals by introducing an incentive system. By introducing an incentive system the firm can set targets on the factors that matter the most to the firm. By doing so, the employee will focus on these targets because there is an incentive attached to making that target. As a result, the firm and the employee have the same objectives. The second problem that occurs is the risk-sharing problem as discussed by Arrow (1971) and Wilson (1968) in which the principle and agent have different risk preferences and therefore make different decisions. The assumption is that the agent is more risk averse than the principle (Eisenhardt, 1989). When the agent receives a higher reward he will accept a higher risk. In this way, the principle and agent share the risk and align their risk preferences. The main focus of agency theory is finding the most efficient contract governing the principle-agent relationship (Eisenhardt, 1989). This means choosing the contract with the highest productivity of the agent for the lowest costs.
Incentive systems and gaming
The power of incentive systems and the effect on employee behavior and firm performance has been excessively discussed in the literature (Baker et al., 1987; Holmstrom, 1979; Levinthal, 1988). Incentive systems are designed to align the objectives of the firm and the firm’s employees (Ethiraj & Levinthal, 2009; Prendergast, 1999) and determine to a great extent how individuals behave in organizations (Baker et. al., 1987). Positive effects of incentive systems are that they motivate employees (Holmstrom, 1979; Levinthal, 1988),
enhance productivity (Ichniowsky, Shaw, and Prennushi, 1997; Obloj & Sengul, 2012) and attract high ability employees (Clinch, 1991; Levinthal, 1988). The literature also pays attention to the negative effects of incentive systems and how these negative effects can harm instead of benefit the organization (Kreps, 1997; Larkin et. al., 2012; Ordonez et. al., 2009). According to Ordonez et. al. (2009) incentives work counterproductively because they shift the focus away from important but not specified goals, harm interpersonal relationships, corrode organizational culture, and motivate risky and unethical behaviors.
Obloj and Sengul (2012) argue that there are two types of learning that explain the positive and negative effects of incentive systems. Productive learning leads to employees learning over time and with experience how to be better in a task, which enhances productivity. Adverse learning leads to negative side effects of incentives systems, because employees learn how to exploit the incentive system and how to use it for their personal benefit. When this happens, the main objective of incentive systems to align the firm’s and employees’ interests is not met, because employees pursue their individual goals again instead of the firm’s goals.
Jensen (2003) distinguishes two ways in which employees can exploit incentive systems. They can game the incentive system by either gaming the realization of targets or by gaming the setting of targets. Ways of gaming the realization are the diversion of effort to activities that are rewarded (Larkin, 2014; Oyer, 1998; Pierce, 2012), exploitation of the rules to improve apparent performance (Gravelle, Sutton, and Ma, 2010) and timing (Larkin, 2014; Oyer, 1998).
Although there is a lot of research on the negative side effects of incentive systems and on how to game the incentive system there is not much research on the drivers that lead to gaming. Schweitzer et. al. (2004) found that employees who have a clear goal or are close to achieving a goal are more likely to game. More recently, Frank and Obloj (2014) found that high firm-specific human capital managers are more productive in their primary task but are also more likely to game. In the next sections other drivers of gaming will be identified.
Tenure
Frank and Obloj (2014) identified a link between firm-specific human capital and gaming. When firm-specific human capital is high, it leads to higher productivity but also to more gaming. Because of advanced skills and knowledge, employees with high firm-specific human capital are becoming more productive, but are also getting better at gaming. These findings might also indicate a relation between tenure and gaming. Employees who are in the firm for a longer period of time will learn more about the company and the incentive system. They can use this knowledge to their own advantage by gaming the incentive system.
Following the theory on productive learning (Obloj & Sengul, 2012), employees will get better at their job over time. Therefore, it seems logical that when being in the company for a longer period of time, employees will have more time and effort left to spend on “cracking the incentive system”. These arguments results into the following hypothesis:
H1: Tenure is positively related to gaming incentive systems.
The relation between tenure and timing gaming has been previously examined by Larkin (2014). However, even though Larkin (2014) found an interaction between tenure and
to gaming or enjoy gaming stay with the company. From another perspective, it could also be that employees who are working for the same company for a while feel connected with the company and therefore do not want to act unethically out of loyalty. Therefore, loyalty is expected to have a moderating effect on the relation between gaming and tenure. This results in the following hypothesis:
H2: Employee loyalty has a negatively moderating effect on the relation of tenure and gaming incentive systems.
Unfairness - wealth-based inequity
Unethical and dishonest behavior has been excessively discussed over the years. In 1999, Fehr and Schmidt defined unfairness as an aversion of inequity. Larkin and Pierce (2015) describe how unfairness reduces employee effort. They identify four factors that establish perceived fairness: comparison to reference points, comparison to others, minimum wage and prior income. Recently, the literature has focused on comparison to others and how this leads to perceived wealth-based inequity which, in turn, leads to unethical behavior (Gino & Pierce, 2009a, 2009b, 2010; John et al., 2014). In their experiments, Gino and Pierce (2009a, 2009b, 2010) found that dishonesty is influenced by emotional reactions to wealth-based inequity. They found that negative inequity leads to feelings of envy and results in hurting behavior whereas positive inequity leads to feelings of guilt and results in helping behavior. Sometimes dishonesty can lead to personal financial costs, but even in that case wealth-based inequity causes dishonesty. Following these findings, John et al. (2014) found that dishonesty depends on how people can compare their pay-rate. Employees can be aware of the pay-rate of
colleagues or peers within or across firms. This can lead to wealth-based inequity when an employee finds out a colleague or peer has a higher pay-rate. Another factor that induces the
feeling of inequity is employee overconfidence. Employees tend to overestimate their skills and performance and as a result perceive the incentive system as unfair (Larkin et al., 2012). When overestimating their own abilities, the feeling of unfairness and inequity grows. This growth occurs not only when you know a colleague or peer has a higher pay-rate, but also when you do not know the pay-rates or when the pay-rate is equal. When employees
overestimate their own performance or abilities, they feel like they are better at their job than colleagues or peers and therefore should get a higher reward. As a result they have a feeling of unfairness and wealth-based inequity. Following the literature, this can result in dishonesty or unethical behavior. As a form of unethical behavior, gaming the incentive system should be induced by a higher perceived wealth-based inequity. This results in the following
hypotheses:
H3: Perceived wealth-based inequity is positively related to gaming incentive systems.
H4: Employee overconfidence has a positively moderating effect on the relation of perceived wealth-based inequity and gaming incentive systems.
Personal wealth
Following the literature it seems that tenure and wealth-based inequity are drivers for
unethical behavior and therewith for gaming incentive systems. The reason for this would be the feeling of being treated unfairly. Following this idea of unfair treatment, personal wealth, in a sense of how much money a person has, could also have an influence on gaming. Low personal wealth might lead to feelings of unfairness because it might feel unfair to have low personal wealth while others are wealthier for no apparent reason. Therefore it seems possible that employees might want to compensate for their “bad luck” by compensating their low
personal wealth by gaming. Aside from unfairness, it seems logical that a person who needs money would tempted to game sooner than someone who has enough money. The main assumption of agency theory is that people are self-interested which means they have and pursue their individual needs (Eisenhardt, 1989). This means that people are only motivated by their personal monetary payoffs. The idea is that people think that money is good and effort is bad. Thus, they will try to make the most money with the least effort. Following this idea, it can be argued that people with low personal wealth will try to improve their wealth by gaming the incentive system. A form of gaming is focusing on activities that are rewarded and not on activities that are not rewarded (Larkin, 2014; Pierce, 2012; Oyer, 1998), hence
limiting effort for the highest reward. This fits the idea of agency theory that money is good and effort is bad. People with low personal wealth might use this method more often because they need it more. The need for self-interest is higher. This results in the following
hypothesis:
H5a: When an agent’s perceived personal wealth is low it induces gaming incentive systems.
Aside from arguing that low personal wealth is a driver to gaming it can also be argued that high personal wealth is a driver for gaming. Following agency theory, people are
self-interested (Eisenhardt, 1989) and are only self-interested in money and personal gain. Therefore, it could also be that the more financial resources you have, the more you need or want. This results in the final hypothesis:
H5b: When an agent’s perceived personal wealth is high it induces gaming incentive systems.
Visualization
The theory discussed and the hypotheses following this theory can be summarized as follows (see figure 1):
Figure 1. Conceptual model.
METHODOLOGY Research design
This paper aims to study whether tenure, wealth-based inequity and personal wealth are factors in driving employees into gaming incentive systems by answering the following research question: How do tenure, wealth-based inequity and personal wealth influence gaming incentive systems? In an attempt to answer the research question and hypotheses, an experiment will be conducted. Gaming incentive systems is a delicate subject, because it can
be seen as employees behaving unethically or doing something that is not allowed. Therefore, it will be difficult to use methods like interviews or surveys because it is likely that most participants will not answer honestly when asked about their gaming behavior. An experiment makes it possible to create specific circumstances in order to find out more about gaming behavior under these specific conditions. This will, hopefully, lead to honest responses to incentive systems under the created conditions. This research is a mixed-method research, because it involves an experiment that also includes surveys. There will be questionnaires during the experiment to make sure that the participants perceive the manipulations as
intended. For example, in the experiment, salary is reduced to create a feeling of low personal wealth. The questionnaire after this manipulation is expected to show a lower score on
personal wealth than the questionnaire before the manipulation.
Level of analysis
This research is conducted at the individual level. Both the unit of analysis and level of analysis are at the individual level, which means that data is collected from individuals and conclusions are drawn over individuals.
Sample
The population of this research is defined as: “every adult person in the world that is not in a top management position”. There is no access to a sampling frame, which makes it impossible to use probability sampling. The experiment will take approximately 1,5 hours and requires participants to come to a specific location. Therefore a condition for the sample is that participants are in the Netherlands. Time also makes it more difficult to get people to participate in the experiment than, for example, a survey that takes only 15 minutes and can be filled out from your own computer at home. Therefore, the size of the population and time
lead this research to non-probability sampling. Two forms of non-probability sampling will be used. First, convenience sampling is used which means approaching participants who are most convenient to approach. This will include colleagues, friends and family. These participants will be approached in person or by telephone by the researcher. Second, these participants are asked to identify further participants, which is known as snowball sampling. These participants will be approached by the first group of participants in person or by telephone. The personal approach is expected to lead to a high response rate. There are a few requirements for the participants. First, participants should be in the Netherlands, because the experiment takes place in the Netherlands. Second, they should be, in regards of hierarchy, positioned below the top management. Finally, participants need to be able to understand a working situation. Therefore the targets are defined as: “every adult person in the Netherlands that is, in regards of hierarchy, positioned below the top management”. There are two groups in the experiment: the control group and the experimental group. There are multiple variables tested in this research, and the control and experimental conditions switch between the two groups. The two groups are designed to be demographically as equal as possible. To accomplish this, participants are asked to fill out a survey with demographical questions before the experiment. Based on this survey, participants are divided into two equal groups.
Research procedure
Data is collected through the conduction of an experiment. Figure 2 shows an overview of the phases in the experiment.
Figure 2: planning experiment
The experiment starts with the pre-experiment phase. In this phase, participants are e-mailed an URL to an online form to register. This form asks for participants’ name, e-mail address, gender, age, education level and nationality. It is noted that the information is only used for the administration of the experiment. The registered participants are divided into two groups. The goal is to have two groups that are demographically equal. To accomplish equality between the groups the distribution of demographic aspects has to be equal. First, groups are randomly assigned by ordering the participants names from A to Z. It is expected that the groups will not be equal after this, so groups are adjusted by switching participants from group to group until the groups are equal. The second phase of the experiment is the actual experiment. The experiments of the two groups are not conducted at the same time due to
and therefore the experiment has multiple rounds. The control and experimental conditions switch between the two groups to avoid boredom for the control group. For example, in the first round tenure and loyalty are tested with group 1 as the control group and group 2 as the experimental group. In the second round, high personal wealth is tested and group 2 will be the control group and group 1 the experimental group. In the third round, low personal wealth is tested and group 1 will be the control group again. There are multiple independent variables that are tested in this research (tenure, loyalty, personal wealth, wealth-based inequity and overconfidence). Therefore, there are several manipulations that need to be reversed in order to test the next variable. In order to let participants forget the previous manipulation there will be a break after the manipulations so it will be clear that a new phase of the experiment is starting. The experiment starts with a general instruction. There will be a PowerPoint
presentation to lead this to make sure all participants get the same instruction. Participants are told to listen carefully to what is told during the experiment and think about what this
situation means for them and how they would feel about it. It is important for participants to behave the way they would behave in the given situation. During the experiment it is not allowed to talk with other participants. After this brief introduction, the starting situation is explained to all participants. They are told they are working for “Puzzles”, a company specialized in solving puzzles. The more puzzles they solve in each month, the better this is for the company. One month lasts 2 minutes. After every month participants are paid a basic salary of 10 euros. For every puzzle they solve from 50% of their target they receive 0,50 euro bonus. For example, if a participant has a target of 6 and solves 6 puzzles he receives 11,50 euro at the end of the month. Following the introduction, there will be one practice month for both groups to get an idea of the type of puzzle. After this, the experiment starts with tenure and loyalty (group 1 control, group 2 experimental), followed by personal wealth high (group 1 experimental, group 2 control), personal wealth low (group 1 control, group 2
experimental), wealth-based inequity and overconfidence (group 1 experimental, group 2 control). These factors will be manipulated for the experimental group. The last phase of the experiment is the post-experiment phase where participants are thanked for their participation. Data analysis will follow.
Variables
Dependent variables
The dependent variable that is measured in this research is gaming of the incentive system. Gaming the incentive system is defined as: exploitation of the incentive system for the personal benefit of the employee in a way that hurts the organization. In this research, it is possible to game the setting of targets and to game the realization of targets. Participants are asked to write down their own target before the start of every month. This gives them an opportunity to game their target by setting a low target. Participants know from the practice month how many puzzles they are able to solve within a month. So the question here is whether they write down a realistic target, or game the system by writing down a lower target in order to receive a higher incentive. Another way to game the incentive system in this research is by gaming the realization of targets. Participants are asked to write down how many puzzles they solved at the end of each month. However, the reported amount is not checked. Therefore, there is an opportunity to game by writing down a different number of solved puzzles than actually solved in order to receive a higher reward. Both types of gaming are measured by checking whether lower targets and/or higher realizations are reported under the different conditions.
Independent variables
There are several independent variables measured in this research. In the experiment, the independent variables will be manipulated for the experimental group. The experiment starts with tenure and loyalty. The first is tenure, which is defined in this research as: the period of time an employee is with the same employer. Tenure is manipulated by giving participants experience with the incentive system (tenure) or by not giving them experience with the incentive system (no tenure). Therefore, gaming is measured after the first month in the experiment, when participants have no experience with the system and for example do not know that they can set their own target. Before the first month, participants are asked to write down how many puzzles they think they can solve in the first month. After writing this down, they are told that the number they wrote down will be their target. In the second and third month, they already know that they can write down their own target because of their
experience with the system. Will this lead to more gaming? The same happens with reporting the amount of puzzles solved. Participants are asked to write down the amount of puzzles they solved at the end of each month. In the first month, participants have no experience with the system and might expect that the amount of puzzles they report as solved will be checked. In the second and third month, they already experienced that this is not the case. This might lead to more gaming. Loyalty is expected to be a moderating variable of tenure and gaming. Loyalty is defined in this research as: the commitment and the connection that an employee feels towards their employer. It is measured by adding something to the story of one of the groups, which is expected to increase loyalty. In the first month of the experiment, the no tenure month, the control group gets no information about loyalty. The experimental group gets the story that even though they have been working at Puzzles for a short period of time, they immediately feel connected to the company. This is intended to increase loyalty. In the second and third month of the experiment, the tenure months, participants in the experimental
group are told that they are already working at Puzzles for 25 years now. To celebrate this, their boss organized a small party thanking them for their good work and honesty. Will there be less gaming in the group that has tenure and loyalty compared to the group that has tenure but no loyalty? Figure 3 shows the design of the experiment for tenure and loyalty.
Figure 3: measurement of tenure and loyalty
Participants should be reset after the third month, as after this loyalty cannot play a role in their decision to game. Therefore, participants are told before the fourth month that they can forget about the amount of years they have worked at Puzzles. They can now assume they have worked at Puzzles for a couple years and feel normally involved and connected. To make sure the loyalty group forgets about this, there is a break between month three and four. The second independent variable in this research is perceived personal wealth. In this
research, perceived personal wealth is defined as: the perceived extent of accessible money. Until this moment in the experiment, all participants received a basic salary of 10 euros per month. Participants are told at the start of the experiment that this is a normal amount from which they can live normally without problems. High personal wealth is manipulated by increasing the salary of the experimental group to 50 euros per month. This is expected to give them the feeling of high personal wealth, because this is a lot higher than their salary so far from which they could live normally. To make sure the manipulation works participants are told that 50 euros is a salary from which they can live extremely well. The control group keeps a salary of 10 euros per month. In the next round, low personal wealth is manipulated by reducing the fixed salary of the experimental group to 1 euro per month. This is expected
to give a feeling of low personal wealth because this is a lot lower than their salary so far (10 euros per month). To make sure the manipulation works, participants are told that this reduction means that it will be harder for them to make it through the month. The control group, which was the experimental group in the high personal wealth month, gets a salary of 10 euros per month. There is a break in between the high and low personal wealth round for this group because the purpose is not to make them feel like they received a salary reduction. There is a break to make them forget the high personal wealth manipulation. To make sure this works, they are told at the beginning of the low personal wealth month that the salary increase unfortunately was just a dream. It did not occur and therefore the switch back to a 10 euro salary is no reduction. Figure 4 shows the experimental design for perceived personal wealth.
Figure 4: measurement of personal wealth
After the fifth month, the experimental group for low personal wealth needs to be reset. They are told the salary reduction was just a nightmare and did not occur. To make sure they forget about the reduction there is a break between month five and six for this group. The last independent variable in this research is perceived wealth-based inequity with a moderating effect of overconfidence. In the sixth month, overconfidence is measured between the groups. Overconfidence is defined in this research as: the perception to be better at something than your peers. It is manipulated by telling participants in the experimental group that they just had their review and they got the news that they perform extremely well. They belong to the top puzzle-solvers of the company. This is intended to increase the feeling to be better than
colleagues. The control group also gets the story of the review, but is told that all puzzle-solvers at Puzzles are at the same level. This is intended to make participants feel like they are just as good as their colleagues. In this month, there is no inequity. To make sure participants perceive this as such, both groups are told all puzzle-solvers earn the same amount of money (10 euros). The within-groups factor in month six and seven is inequity, which is defined in this research as: perceiving lower personal wealth compared to other people that are perceived to be our equals. Wealth-based inequity is manipulated in the seventh month by telling the participants in the overconfidence group that colleagues with the same job and capabilities are getting paid a higher salary than they are, even though they are way better at solving the puzzles. This is expected to create a feeling of inequity. The group that has no overconfidence is told that even though all puzzle-solvers have the same skills, there are colleagues with a higher salary. This is expected to create a feeling of inequity as well. Figure 5 shows the experimental design for perceived wealth-based inequity and overconfidence.
Figure 5: measurement of wealth-based inequity and overconfidence
Data analysis
After conducting the experiment, analysis of the data will follow. This starts with preparing the data for analysis, followed by preliminary analysis, selecting an analysis method, manipulation check and the actual analysis. First, the data is prepared for the analysis in SPSS. This mostly means coding the data by, for example, putting a 1 for male and a 2 for female. Also, the dependent variable gaming needs to be transformed into one outcome. Gaming is not measured as one outcome, but there are several measures that together express
gaming. Both target and realization can be gamed by the participant by setting a low target or over-reporting realization. This leads to three results: the target the participant set, the
reported realization (reported amount of puzzles solved) and the actual realization (real amount of puzzles solved). These results can be transformed into one outcome for gaming of the target and one outcome for gaming of the realization. Gaming of realization is then calculated as follows:
Gaming of realization = reported realization of current month – actual realization of current month
A quick look at the collected data showed that there was no gaming of the realization. All participants reported the actual amount of puzzles solved and did not over report their performance. Gaming of targets is not that easily checked because you only have the target the participant wrote down and no other number to check this. Logically thinking every participant should be able to solve as many puzzles in one month as the month before. Therefore, it would make sense if the participant would write down the amount of puzzles solved in the previous month as a target for the next month. Because there is no gaming of the realization, this way of computing the outcome of gaming targets is possible.
Gaming of target = target the participant set – actual realization of the previous month
The data is now prepared and preliminary analysis will follow next. This will involve checking the descriptive statistics, assessing normality and checking for outliers. This research involves several variables that are tested on two groups (control and experimental group). Some of the variables are tested within-groups and others are tested between the
groups. Therefore, several statistical methods are needed to analyze the data. The first independent variable is tenure and is measured within the groups. In the first month there is no tenure and in the third month there is tenure. Loyalty is expected to have a moderating effect on the relation between tenure and gaming. Loyalty is tested between the groups, as one group has loyalty and the other group does not. Thus, there is a mixed between/within-subject design and two-way mixed ANOVA will be used to analyze the results. The first three
assumptions for a two-way mixed ANOVA all fit the design and measurement of this research: there is one continuous dependent variable (gaming), there is one between-subject variable (loyalty) that has two categorical levels (loyalty and no loyalty), and there is one within-subjects variable (tenure) that has to categorical levels (no tenure and tenure). The main reason for choosing two-way mixed ANOVA is that it allows you to establish whether there is an interaction effect between the between-subjects and within-subjects factor. When the interaction is significant a simple main effects analysis will follow and when the
interaction is not significant a main effects analysis follows. Because there are only two groups, it is not necessary to perform post-hoc tests (Pallant, 2013). The second independent variable is personal wealth and is measured on two levels: high personal wealth and low personal wealth. These are tested between the groups and there is no moderator. The objective is to test the difference between the two groups. Therefore, an independent sample T-test is performed for both high personal wealth and low personal wealth. There are only two groups and therefore the independent sample T-test is chosen over one-way ANOVA. The last independent variable tested is wealth-based inequity with an expected moderating effect of overconfidence. Wealth-based inequity is measured within the groups and overconfidence is measured between the groups. A mixed ANOVA will be performed in line with tenure and loyalty. Again, there are only two groups and therefore a post-hoc test is not necessary. Before performing the selected analysis methods, there will be a check of the manipulations.
Did the participants receive all manipulations as intended? All participants filled out a questionnaire during the experiment to check whether the manipulations were received as such. For example, in the personal wealth high month, the experimental group should have a significantly higher score on personal wealth than the control group because of the
manipulation. Before performing the selected analysis methods, there will first be an analysis of the questionnaires. When a question checking the manipulation does not show a significant difference between the control and experimental group, the data cannot be further analyzed. The questionnaire checks the manipulations of loyalty, personal wealth, wealth-based inequity and overconfidence. Tenure is not checked, since it is difficult to measure this in a question and because there is a clear difference in experience after the first and third month. An
independent sample T-test is used to check the between-group manipulations loyalty, personal wealth and overconfidence. A paired-sample T-test is used to check the within-group
manipulation of wealth-based inequity. After the manipulation check, the actual analysis will follow.
RESULTS
The main goal of this research is to prove whether tenure, perceived wealth-based inequity and perceived personal wealth cause gaming incentive systems. The manipulations of these independent variables have been checked in the experiment through questionnaires after each month. This is done in order to check whether the participants received the manipulation as intended. The results of the manipulation check are shown in appendix 1. All manipulations showed a significant difference between (loyalty, personal wealth, overconfidence) and within (inequity) the groups (p < .05). The participants received all manipulations as intended, which means all independent variables can be tested.
This research performs independent sample T-tests, paired sample T-tests and two-way mixed ANOVA to test the hypotheses. The two-way mixed ANOVA has assumptions for outliers, normality, covariance, variance and sphericity. The T-tests have assumptions for outliers, normality and homogeneity of variances. As the first three assumptions are the same for all tests, they will be checked first for all hypotheses. The results are presented in appendices 2 to 4.There were three outliers in the data, which had studentized residual value of >3. All three outliers have been removed from the data. Shapiro-Wilk’s test shows that gaming is not normally distributed for all groups on all independent variables (p < .05). It is not possible to transform the data, because then all between and within-groups need to be transformed. Not all groups have the same distribution and therefore it is not possible to transform them. This is why the data is not transformed and it is accepted in this research to violate the normality assumption. There are two variables that do not have homogeneity of variances, as assessed by Levene’s test of homogeneity of variance (p < .05). One of the variables is perceived personal wealth low (p = .002) for which an independent samples T-test is performed. It is no issue to violate this assumption because it just means to use the equal variances not assumed column. The other variable that has no equal variances is tenure (p = .047). It is not possible to transform the data, so the violation is accepted. All assumptions for the independent samples T-test are discussed. The last two assumptions are for the first two hypotheses (tenure) and second two hypotheses (wealth-based inequity) that will be tested with the two-way mixed ANOVA.Both tenure and wealth-based inequity have homogeneity of
covariances, as assessed by Box’s test of equality of covariance matrices (p = .517 (tenure) and p = .062 (perceived wealth-based inequity). The final assumption for two-way mixed ANOVA is met, because Mauchly’s test of sphericity indicated that the assumption of sphericity was met for the two-way interaction (p > .05). All assumptions have now been checked.
The results of the first two hypotheses are presented in tables 1 and 2. The descriptive statistics and assumption tests for these hypotheses can be found in appendix 2. First, the interaction effect is explored and the results of this effect which relate to hypothesis 2 are presented in table 1. There is no statistically significant interaction between tenure and loyalty on gaming, F(1, 77) = 1.172, p = .282, partial η2 = .015. These results mean that there is no support for hypothesis 2, that loyalty decreases the effect of tenure on gaming incentive systems.
Test of within subjects effects F df Sig. Partial eta squared
Tenure * group 1.172 1 .282 .015
Table 1: results of hypothesis 2
Hypothesis 1 is on the effect of tenure alone. The results are presented in table 2. The main effect of tenure shows a statistically significant difference in gaming with or without tenure,
F(1, 77) = 39.985, p <.0005, partial η2 = .342. These results provide strong support for hypothesis 1, that tenure induces gaming incentive systems. The effect size (partial η2 = .342) indicates a strong effect. Although this test is two-sided while the hypothesis is one-sided, the descriptive statistics show the direction of the effect. The mean of the month without tenure (mean = 0.0759) is higher than the mean of the month with tenure (mean = -1.9557). This means there is more gaming in the tenure month, which shows the direction of the effect is the same as in the hypothesis.
Test of within subjects effects F df Sig. Partial eta squared
Tenure 39.985 1 .000 .342
Table 2: results of hypothesis 1
Although there is no hypothesis for the effect of loyalty alone, this effect is explored as well. The results can be found in table 3. The main effect of loyalty shows no statistically
significant difference in gaming with or without loyalty, F(1, 77) = .765, p = .384, partial η2 = .010. This means loyalty by itself has no effect on gaming incentive systems.
Test of between subjects effects F df Sig. Partial eta squared
Loyalty .765 1 .384 .100
Table 3: results of loyalty effect
The results of the second two hypotheses can be found in tables 2 and 3. The descriptive statistics and assumption tests of these hypotheses are presented in appendix 3. First the interaction effect is explored to test hypothesis 4. The results are presented in table 4. There is a statistically significant interaction between perceived wealth-based inequity and
overconfidence on gaming, F(1, 77) = 6.443, p = .013, partial η2 = .077. These results mean that there is support for hypothesis 4, that employee overconfidence increases gaming incentive systems caused by perceived wealth-based inequity. The effect size (partial η2 = .077) indicates a medium effect.
Test of within subjects effects F df Sig. Partial eta squared
Inequity * group 6.443 1 .013 .077
Table 4: results hypothesis 4
There is a significant interaction and therefore simple main effects are explored next. The results of hypothesis 3 are presented in table 5. The simple main effect of wealth-based inequity showed a statistically significant difference in gaming with or without wealth-based inequity when there was no overconfidence, F(1, 39) = 32.126, p < .0005, partial η2 = .452. The simple main effect of wealth-based inequity showed a statistically significant difference in gaming with or without wealth-based inequity when there was overconfidence, F(1, 38) = 15.862, p < .0005, partial η2 = .294. These results provide strong support for hypothesis 3, that perceived wealth-based inequity induces gaming incentive systems. Although this test is
two-sided while the hypothesis is one-sided, the descriptive statistics show the results go in the same direction as the hypothesis. Both groups (with and without overconfidence) show more gaming when there is perceived wealth-based inequity (µ (group 1) = -6.4438 and µ (group 2) = 5.7949) than when there is no perceived wealthbased inequity (µ (group 1) = -3.1375 and µ (group 2) = -4.2692).
Test of within subjects effects F df Sig. Partial eta squared
Group 1 32.126 1 .000 .452
Group 2 15.862 1 .000 .294
Table 5: results hypothesis 3
Although there is no hypothesis for the effect of overconfidence alone the results are presented in table 6. The simple main effect of overconfidence shows no statistically
significant difference in gaming with or without overconfidence when there was no inequity,
F(1, 77) = 2.074, p = .154, partial η2 = .026. The simple main effect of overconfidence shows no statistically significant difference in gaming with or without overconfidence when there was inequity, F(1, 77) = .404, p = .527, partial η2 = .005. This means overconfidence on itself has no effect on gaming incentive systems.
Test of between subjects effects F df Sig. Partial eta squared
No inequity 2.074 1 .154 .026
Inequity .404 1 .527 .005
Table 6: results of overconfidence effect
An independent sample T-test is used for hypotheses 5a and 5b. The descriptive statistics and results of the assumption tests can be found in appendix 4. The results of hypothesis 5a can be found in table 7. Hypothesis 5a states that perceived low personal wealth induces gaming. Data are mean±standard deviation, unless otherwise stated. The results show that there was more gaming in the low personal wealth group (-6.45±4.87) than in the control group (-3.64
test for equality of variances (p = .035). There is a statistically significant difference in gaming between the experimental group (perceived low personal wealth) and the control group, t(77) = -3.083, p = .003. These results provide support for hypothesis 5a, that perceived low personal wealth induces gaming incentive systems.
Variables t df Sig. MD SDE
Gaming perceived low personal wealth -3.083 65.622 .003 -2.8090 .9111
Table 7: results hypothesis 5a
The results of hypothesis 5b are presented in table 8. Hypothesis 5b states that perceived high personal wealth induces gaming. The results show that there was less gaming in the high personal wealth group (-2.00±3.28) than in the control group (-3.08±3.04). There was homogeneity of variances for gaming for high personal wealth and the control group, as assessed by Levene’s test for equality of variances (p = .801). Gaming with perceived high personal wealth was -1.08 (95% CI, -2.50 to .33) lower than gaming in the control group. There is no statistically significant difference in gaming between the experimental (perceived high personal wealth) and control group, t(77) = -1.522, p = .132. These results provide no support for hypothesis 5b, that perceived high personal wealth induces gaming incentive systems.
Variables t df Sig. MD SDE
Gaming perceived high personal wealth -1.522 77 .132 -1.0813 .7105
Table 8: results hypothesis 5b
CONCLUSION AND DISCUSSION
The goal of this research is to study the effect of tenure, perceived wealth-based inequity and perceived personal wealth on gaming incentive systems. Hypothesis 1 states that there is a positive relation between tenure and employees gaming incentive systems. The results show
support for this hypothesis, which is in line with theories of Obloj & Sengul (2012) and Frank and Obloj (2014) on adverse learning and firm-specific human capital. This research shows that, indeed, employees who have experience with the incentive system will exploit it for their personal benefit. In the first month of the experiment, none of the participants are told they can set their own target or that the amount of solved puzzles is not checked. Then, after the first month, participants learned from experience that this was, in fact, the case. This immediately led to more gaming. Obloj & Sengul (2012) suggested that both productive learning and adverse learning lead to gaming. This research shows that adverse learning rather than productive learning leads to gaming, as gaming was not so much caused by the extra time participants had to think about how to “crack the system”, but rather seemed to come more from experience with the incentive system and the skills and knowledge employees develop about it. Participants did get more productive during the experiment, which was shown because the amount of solved puzzles went up during the experiment. This shows that people do not use the extra time to think about the system, but started gaming because of the extra knowledge they gained about the system.
The second hypothesis claims that loyalty has a negative moderating effect on the relation of tenure and gaming. No support has been found in this research for this hypothesis. The
relation between only loyalty and gaming was also explored in this research, but did not show an effect. Earlier, Larkin (2014) found a relation between tenure and gaming in the form of timing, but it remained unclear whether employees learned to game over time or if those who are used to gaming or enjoy gaming stay with the company. Not finding an effect of loyalty on gaming nor a moderating effect of loyalty on tenure and gaming could indicate that employees who like to game or are used to gaming stay with the company and therefore loyalty has no effect on gaming. However, it was not possible to quit the company in this
research so this cannot be proven. Future research could look into whether employees who are used to or enjoy gaming stay within the company and whether there is a link or interaction with loyalty. There is an indication in this research that employees learn to game over time. This research does not only show a difference in gaming with or without tenure, but it also shows that gaming increases throughout the months for both the control and experimental group. Thus, apart from the other manipulations in this research, gaming increases due to learning, which is in line with the theory of Obloj and Sengul (2012). Another reason for not finding an effect of loyalty on tenure might be that it is hard to create a true feeling of loyalty in an experiment. The assumption is that employees who feel more connected to the company would not feel good about themselves when gaming and would therefore not game at all, or at least to a lower degree. It is difficult to give participants the true feeling someone has towards an employer after 25 years. This is a limitation to this research. Also, in general, an
experiment can feel like a game to some participants, which can influence the results. Therefore, a suggestion for further research is to look into the role of loyalty in another way than through an experiment.
Hypothesis 3 claims that perceived wealth-based inequity is positively related to gaming incentive systems. The results show support for this hypothesis. Larkin and Pierce (2015) identified comparison to others as a factor that establishes perceived fairness. This research showed that gaming was higher when people knew there was inequity in salary and they were earning less than others. This indicates that people do not perceive it as fair when there is inequity in salary and this leads them to gaming. This is also in line with unethical behavior that comes from perceived wealth-based inequity as described earlier by Gino & Pierce (2009a, 2009b, 2010) and John et al (2014). In 1999, Fehr and Schmidt already defined unfairness as an aversion of inequity. This research shows that, indeed, perceived
wealth-
based inequity gives employees a feeling of unfairness, because they distance themselves from the system by gaming it. Larkin and Pierce (2015) suggested that perceived unfairness leads to less employee effort, which did not show in this research. Productivity went up, and so did gaming. It is interesting that there was no gaming of the realization in this research and that productivity went up. These findings might indicate that employees see target gaming as “being smart” rather than doing something that is not right or allowed. Working less hard or gaming realization could be perceived as such. Future research could look into the perception of people on different types of gaming and which forms of gaming are perceived as wrong. The results also indicate that people take matters into their own hands by compensating their wealth by gaming. According to Gino and Pierce (2009a, 2009b, 2010), dishonesty is
influenced by emotional reactions to based inequity. An emotional reaction to wealth-based inequity could be to compensate for the dishonesty of earning less money than others by gaming the system in order to receive more salary. This research does not look into the underlying emotions of people, which is a limitation. Further research could look into the underlying emotions of people that result into gaming. This research also does not look into hurting and helping behavior that comes from perceived wealth-based inequity as described by Gino & Pierce (2009a, 2009b, 2010) and John et al (2014). When a participant games in this research, it will only benefit himself. Future research could look into hurting and helping behavior in incentive systems and see how people respond when their actions would actually hurt or benefit not only themselves but others as well.
The fourth hypothesis predicts a positive moderating effect of overconfidence on the relation of perceived wealth-based inequity and gaming. The results support this hypothesis. Larkin et al (2012) found that people who overestimate their skills and performance perceive the incentive system as unfair. They also suggested that the feeling of unfairness not only grows
when employees know the rates of others, but also when employees do not know the pay-rates or when the pay-rate is equal. This suggests that overconfident employees game with or without perceived wealth-based inequity. This research looked into the differences in gaming when there was no inequity and the results showed that when the pay-rate was equal and there was overconfidence there was no difference in gaming than when there was no
overconfidence. These findings go against the theory of Larkin et al (2012). This research also showed no differences in gaming with or without overconfidence when there was perceived wealth-based inequity. These results show that overconfidence has no direct effect on gaming. However, there was found a moderating effect of overconfidence on the effect of wealth-based inequity on gaming. This means that even though overconfidence alone has no direct effect on gaming, it does strengthen the effect of perceived wealth-based inequity on gaming. Hence overconfidence alone is not enough reason to start gaming and it is not perceived as unfair, or at least not enough to start gaming. However, when combined with perceived wealth-based inequity it is perceived as unfair. This implies that people do not feel like overconfidence alone is enough reason to start gaming. Future research could look into the reason why overconfidence alone is not enough reason to start gaming, but does strengthen the effect of perceived wealth-based inequity on gaming. The results of overconfidence on gaming could also be influenced by the way of manipulating overconfidence. This research gave people overconfidence by telling them they were better than others. It could be that people who are overconfident have specific characteristics or are, for example, always in the less performing group. By giving everyone overconfidence these characteristics are ignored. This could have led to different results and is a limitation to this research. Future research could look into this more.
Hypothesis 5a states that perceived low personal wealth has a positive effect on gaming. This research found support for this hypothesis. According to agency theory (Eisenhardt, 1989) people are self-interested and therefore only pursue their individual needs. Following this theory, it can be argued that people with perceived low personal wealth have the personal need to make more money to increase their personal wealth and therewith fulfill their
individual need. The results of this research show that, indeed, perceived low personal wealth causes gaming, but what is the underlying motivation? Is this caused by the assumption of agency theory that people are self-interested? Agency theory also states that people want to make the most money with the least effort (Eisenhardt, 1989). If this were true, it is expected that people with low perceived personal wealth would game a lot in order to earn the most money with the least effort. The results show that productivity did not go up or down when personal wealth was low. Gaming however did increase under low personal wealth. This means that people put in the same amount of effort as before, but still want to make more money and therefore game more. This could indicate that agency theory is right and that individual need comes first. Another underlying motivation could be unfairness. This research shows that when perceived personal wealth went down, loyalty and perceived wealth-based inequity went down as well. This indicates that perceived low personal wealth leads to a feeling of unfairness, which leads to gaming. Again, this research does not look into these motivations, which is a limitation. Future research could look into this more by exploring the underlying motivations and feelings of people that come from low personal wealth. These results raise questions about the underlying assumptions of incentive systems. The origin of incentive systems lies in the assumptions of agency theory and in the idea that the goals of the principle and the agent do no align (Eisenhardt, 1989). Incentive systems are built on these assumptions, but what if these assumptions do not apply?
The last hypothesis states that perceived high personal wealth has a positive effect on gaming. There was no support for this hypothesis. Following agency theory (Eisenhardt, 1989), the assumption is that people are self-interested and are only interested in money and personal gain. Therefore, you could argue that the more money you have, the more you need or want. The results do not support this theory. They show that gaming went down when salary went up. This implies that there is a limit to the amount of money people need and the extent to which they are only motivated by monetary payoffs. However, the results also show that gaming still happens even though personal wealth is high. This could indicate that there are other possible motivations to game incentive systems, and that they do not have to be monetary. Other possible motivations can be intrinsic motivation or the pressure to perform. This shows again the limitation of this research that underlying motivations are not explained. Future research could look into these other motivations. Agency theory also states that people want to make the most money with the least effort. This does not show in the results.
Productivity did not go down when perceived personal wealth was high. This indicates that people still want to perform even though they do not have to any more because they do not need their bonus. This strengthens the question whether incentive systems are built on the right assumptions. It could also be that the design of the pay structure in this research is of influence here. In the high personal wealth round, salary goes up to 50 euros while the bonus remains 50 cents. The bonus you can earn is not that much in relation to the basic salary. This could have been different when the bonus would go up as well. But even when considering this limitation, the main assumptions of agency theory did not show in this research or need more research to see whether they are true. Incentive systems are built on these assumptions and could therefore have a contrary effect. It is possible that the assumptions of incentive systems are not right and actually lead people to gaming where they would not have gamed when there was no incentive system or the system was build on different assumptions. For
example, targets can lead to pressure to perform, which in turn can cause people to game the incentive system. Thus, incentive systems might be built on wrong assumptions and can actually lead people to gaming and opportunistic behavior instead of preventing it.
MANAGERIAL IMPLICATIONS
There has been a lot of discussion in the literature about incentive systems and the positive and negative effects they bring. Although recently, there is more research on the negative effects of incentive systems, there are still a lot of people and organizations that only see the positive sides of incentive systems and think that the negative effects are not present or do not apply to their own organization. This research shows that, indeed, gaming happens and is caused by several factors. Questions are answered about the effect of tenure, perceived wealth-based inequity and perceived personal wealth on gaming incentive systems.
Hopefully, this research contributes to the acknowledgment of gaming incentive systems in the business and the hazards that it brings.
This research shows that tenure, perceived wealth-based inequity and perceived personal wealth are factors that influence gaming incentive systems. It is important for organizations to realize that these factors exist and to find a way to control them. By ignoring the possibility of gaming or by only seeing the positive sides of incentive systems, organizations might hurt themselves. The factors studied in this research are mostly personal perceptions of the employee. They are being formed by personal experiences. For example, an employee can earn a modal salary and therefore the perception of the employer can be that the personal wealth of this person is fine. However, it could be that this employee has issues, which in return makes his perception of his personal wealth not fine at all. This research shows that low personal wealth is a factor that influences gaming. So, it is possible that the employee
starts gaming based on his own perception. The employer, however, has a different perception and therefore does not expect this. This is why it is important for an employer to understand the factors that lead to gaming and to find a way to control these factors. A way to control these factors could be to get insight in the perceptions of the employee through, for example, one-on-one interviews between manager and employee.
Aside from the negative effects incentive systems have, this research also raises questions about the existence of incentive systems in general and the assumptions on which these incentive systems are designed. Following agency theory (Eisenhardt, 1989), incentive systems are built on the idea that people are self-interested and are only interested in money and personal gain. Organizations claim to need these systems to align people’s personal interests with the firm’s goals. This research shows that there are indications that the assumptions are not entirely true, or at least that there is more to it. Employers should be aware of this and should ask themselves whether the underlying assumptions of their incentive system still apply. When these assumptions turn out to be incorrect, the entire incentive system is built on the wrong assumptions. This is dangerous, because then the system could actually encourage people to being self-interested and behave opportunistic. For example, a firm sets targets in order to align the goals of the employer and the employee. The employee is not self-interested and is not only motivated by money. However, the system is designed on the assumption that he is, so there is a monetary reward when the employee reaches the target, but the employee is also negatively addressed when he does not reach the target. He might even get a bad review for not reaching his target. As a result, the employee feels the pressure to perform. This might encourage him to game in order to reach the target. This is an example on how the incentive system creates the behavior that is assumed but, in
fact, does not exist in the first place. The employee is not only motivated by money, but is assumed to be which eventually encourages him to gaming behavior.
This research raises some interesting questions on incentive systems and their underlying assumptions. Employers should be aware of the dangers incentive systems bring and should ask themselves whether the system still has the wanted effects, or if the system should be redesigned based on different assumptions in order to work. Solutions could be to bring more variation in the systems by, for example, taking intrinsic motivation in consideration, as this could be more important than extrinsic motivation. This research hopes to contribute to this new way of thinking about incentive systems.
REFERENCES
Arrow, K. 1971. Essays in the theory of risk bearing. Chicago: Markham.
Baker, G., Jensen, M., and Murphy. K. 1987. Competition and Incentives: Practice vs Theory,
The Journal of Finance, 43(3): 593-616.
Clinch, G., 1991. Employee compensation and firms’ research and development activity.
Journal of Accounting Research, 29, 59–78.
Eisenhardt, K.M. 1989. Agency theory: An assessment and review. Academy of Management
Review, 14: 57-74.
Ethiraj, S.K., and Levinthal. D. 2009. Hoping for A to Z while rewarding only A: Complex organizations and multiple goals. Organization Science, 20: 4–21.
Fehr, E., Schmidt, K. 1999. A theory of fairness, competition, and cooperation. Quarterly
Journal of Economics, 114: 817–868.
Frank, D.H. and Obloj, T. 2014. Firm-specific human capital, organizational incentives, and agency costs: Evidence from retail banking. Strategic Management Journal, 35(9): 1279-1301.
Gino, F., and Pierce, L. 2009a. 'Dishonesty in the Name of Equity', Psychological Science, 20(9), 1153-1160.
