Punishment or encouragement?
The effect of leadership on individual performance
and free riding under group incentives
Aleksandra Venediktova (11587210)
15 July 18
MSc Business Economics
Track: Managerial Economics and Strategy
Master’s Thesis (15 EC)
Abstract: this thesis is dedicated to the role of leadership in boosting individual performance under group incentives. Many scholars have previously found that leadership can improve coordination in a weak-link game and increase contributions in a public good game. The purpose of this thesis was to compare the effect of two leadership styles, namely encouragement and punishment, on individual choices under group incentives. The scope was further extended to examine the outcomes under group incentives with leadership and in a situation where subjects made individual decisions in the absence of incentives. The results indicate that a leader who strives to encourage the followers to work harder is more likely to increase team performance than the one that intervenes to punish those who deviated from a social norm. Moreover, the motivational effect of encouragement does not extend to a situation when people later get teamed up with strangers with no leadership.
2
Statement of originality
This thesis is written by Aleksandra Venediktova who declares to take full responsibility for its content.
“I declare that the text and work presented in this document are original and no sources other than those mentioned in the text and the references were used in creating this thesis”.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the thesis and not for its contents.
3
Table of Contents
I. Introduction ... 4
II. Related literature ... 5
Group incentives ... 5
Free riding ... 6
Leadership ... 7
III. Methodology ... 10
Experimental design ... 10
Public good game ... 13
Procedures ... 14 Hypotheses ... 17 IV. Results ... 20 Descriptive statistics ... 20 Hypotheses testing ... 21 Leaders’ behavior ... 26
Leadership and free riding ... 28
V. Discussion ... 30 Discussion ... 31 Limitations ... 32 Further research ... 33 VI. Conclusion ... 34 References ... 35 Appendix ... 39
4
I. Introduction
The effect of leadership on motivation and cooperation in groups has been a popular subject of scholarly interest in the past years. In an organization, a leader can play an important role in consolidating the effort of group members and directing it towards the group objective in case its members have different goals that contradict to the main goal of the group (Pogrebna et al., 2009). There is however no univocal opinion about what type of leadership can increase group performance. Some researchers have found that voluntary punishment as a type of leadership can increase contributions in a public good game (Fehr and Gächter, 2000; Fehr and Gächter, 2002; Guth et al., 2007). At the same time, multiple studies conclude that leadership based on encouragement is more likely to enhance group performance as opposed to the one involving punishment (Bass, 1990; Schaumbroeck, Lam and Cha, 2007). However, some researchers conclude that leadership is romanticized and has no impact on team performance (Weber et al., 2001)
This thesis has several features that distinguish it from the existing research and make it an important contribution to the present academic literature on the role of leadership. Firstly, while most academic papers investigate the effect of one form of leadership on behavior in groups (Weber et al., 2001; Moxnes and Van der Heijden, 2003; Brands, Cooper and Fatás, 2007; Güth et al., 2007; Rivas and Sutter, 2011; Cartwright, Gillet and Van Vugt, 2013), this thesis focuses on comparison of two leadership styles on contributions in a variation of a public good game, namely leadership in the form of verbal encouragement and in the form of monitoring and punishment. The results suggest that a leader who encourages employees to act in accordance with group goals can induce higher effort unlike a leader who intervenes to punish those who deviated from a social norm. Secondly, we further analyzed behavior under group incentives by comparing effort choices to a situation when the subjects indicated their preferred allocation of work and leisure time in the absence of monetary incentives. The results indicate that people often prefer to work harder in groups than on their own, which is in line with academic literature on group incentives in the form of profit sharing or employee ownership (Weitzman and Kruse, 1990; Kruse, 1993; Hansen, 1997; Lucifora and Origo, 2015; Bortolotti, Devetag and Ortmann, 2016; Friebel et al., 2017). Moreover, we found that while leadership in the form of encouragement can enhance motivation under group
5
incentives, monitoring does not motivate people to choose an effort above their preferred level. This goes in line with the transactional-transformational theory of leadership (Bass, 1990).
The experiment was designed to test if leadership can motivate people to exert high effort in presence of an opportunity to free ride in a group. The variation of a public good game used in this study was intended to replicate a case when a group of people is working on a group assignment, and their reward is based on group performance. One can think of many instances in which this situation occurs, for example in companies where employee bonuses depend on company performance or in universities where students have group assignments, and everyone in a group receives the same grade. The subjects were asked to choose how they would allocate time between work on a hypothetical assignment and leisure activities. The number of work hours was used as a proxy for individual effort and can be interpreted in terms of a public good game as a contribution to the public account.
The rest of this paper continues as follows. In section 2 we present the review of academic literature on the topics of group incentives, the role of leadership in inducing cooperation and occurrence of free riding in experiments. Section 3 is dedicated to the methodology of the current study. We elaborate on the features of the experimental design, describe the procedures and state the hypotheses. In Section 4 we present the results of the experiment and make conclusions regarding behavior under group incentives and the effect of leadership on effort choices. Section 5 contains discussion and Section 6 concludes the thesis.
II. Related literature
In this chapter, we discuss relevant academic research on the topics of group incentives, the role of leadership in inducing cooperation in public good and weak-link games and occurrence of free riding in laboratory experiments.
Group incentives
A lot of research has been done by both economists and psychologists on the influence of group incentives on individual performance. Generally speaking, there is no consensus
6
regarding whether group incentives have a positive or negative effect on performance of team members and whether group incentives are more efficient compared to individual incentives. The term ‘group incentives’ refers to team-based compensation schemes such as profit sharing or employee ownership plans. Many researchers have found that group incentives in the form of profit sharing or collective performance-related pay are associated with higher productivity compared to individual incentives (Weitzman and Kruse, 1990; Kruse, 1993; Hansen, 1997; Lucifora and Origo, 2015; Bortolotti, Devetag and Ortmann, 2016; Friebel et al., 2017). Group incentives have been found to improve cooperation (Fitzroy and Kraft, 1987; Hansen, 1997; Pendleton, 2006) and promote information sharing within a group (Blasi, Freeman and Kruse, 2016). In an organization, group incentives can signal to the employees that the organization values their input and recognizes their contribution (Park and Kruse, 2014) and thus motivate them to reciprocate by working harder (Eisenberg et al., 2001). Moreover, team-based compensation plans emphasize performance, participation and cooperation and thus enhance the perception of organizational justice, which results in positive employee attitudes and increased performance (Welbourne, Balkin and Gomez-Mejia, 1995; Kwon et al., 2008)
Group incentives induce cooperative values as opposed to individual incentives that are associated with competition within the company (Qin, Johnson and Johndon, 1995). Some scholars conclude that cooperative goals can increase performance due to information sharing and mutual assistance (Wong et al., 2017), while some studies show that competitive compensation is associated with higher performance (Carlson, Upton and Seaman, 2006). The next subsection describes why group incentives can be inferior to individual incentives in their ability to induce optimal outcomes.
Free riding
The problem that is associated with group incentives and which decreases motivational effects of group incentives is free riding (Kandel and Lazear, 1992). Under individual incentives, the rewards reflect the increases in productivity of workers, however under group incentives the gains are shared with other members of the group. The larger the group – the weaker will be the incentives for its members to exert more effort as they will bear the full
7
cost of working harder but they will get only a small share of the increased profits (Baker, Jensen and Murphy, 1988).
Free riding leads to a situation where a Pareto-efficient level of contributions can not be achieved, and the outcome is inefficient. Baker, Jensen and Murphy (1988) conclude that whenever individual performance can be measured, it is always beneficial to use individual incentives and not group incentives to avoid the detrimental effects of free riding. Kandel and Lazear (1992) discuss the free rider problem in partnerships and conclude that occurrence of free riding leads to lower effort compared to the efficient level. In experiments, individual incentives have also been found to be more effective in inducing a socially desired action compared to team-based incentives (Gatiso et al., 2017).
Many scholars have studied cooperation under group incentives employing voluntary contribution experiments (or public good experiments) (Guth et al., 2007; Fischbacher and Gachter, 2010; Rivas and Sutter, 2011). In such experiments, the subjects have an endowment which they can keep to themselves or contribute to the public account. The amount in the public account is then multiplied and distributed evenly among the team members. Despite the fact that in a public good game it is always beneficial to refrain from contributing in order to not only keep the endowment but also receive the proceeds from the public account, in experiments people rarely free ride and usually contribute half of their endowment, which is between a free riding and Pareto efficient levels (Andreoni, 1988). This can be partly explained by the fact that people need time to learn the free rider strategy. Fischbacher and Gachter (2010) found that free riding increases with repetition, which means that there is a learning effect during the experiment. There may be other reasons why people do not free ride in experiments or in real life, such as peer pressure (Kandel and Lazear, 1992), team morale (He, 2012), social preferences such as inequality aversion or altruism (Fehr, Bernhard and Rockenback, 2008), social ties (Harbaugh and Krause, 2000) and so on. In this thesis, we tested if leadership can be the factor that stops people from free riding in a variation of a public good game.
Leadership
Despite the fact that there has been a growing scholarly interest in the effect of leadership on individual performance and team outcomes in the past three decades, existing research is not
8
conclusive about what leadership styles can facilitate organizational performance (Jing and Avery, 2008). Bass (1990) distinguishes between transactional and transformational leadership and states that transformational leadership is related to higher financial performance. Transformational leaders aim at generating awareness and acceptance of purposes and mission of the group in order to motivate followers to look beyond their self-interest, while transactional leadership is associated with contingent rewards and punishment in case the followers deviate from norms and standards (Bass, 1990). Transformational leaders articulate ambitious goals and encourage followers to accept them, which leads to higher team performance (Schaumbroeck, Lam and Cha, 2007).
In experiments, leadership can play an important role in improving coordination in a weak-link game and reducing free riding in a public good game. In a weak-weak-link game, subjects choose a number, and their payoff depends on the minimum chosen in their team. A weak-link game has many Nash equilibria in which people coordinate on the same number (Cartwright, Gillet and Van Vugt, 2013). An example of a weak-link game has been described by Hirschleifer (1983): people living on Anarchia island are responsible for building their section of the wall which will protect them in case of a flood. No matter where the sea enters the island, the damage will be the same for all inhabitants, therefore the level of destruction will depend on the minimal height of the wall. Another example of a weak-link game is given by Weber et al. (2001). A project team is assigned a series of reports that are complete only when all members have finished their sections. Early completion is more costly, however if team members coordinate on an early completion date, they all will receive a bonus.
Due to the fact that in a weak-link game choosing a higher number is costly, achieving a Pareto efficient equilibrium is difficult, which in experiments leads to a coordination failure (Brands, Cooper and Fatás, 2007). Therefore, a leader can improve coordination by choosing a high number before everyone else to signal that this strategy is beneficial for the group and motivate others to select the same. Leadership by example has been found to help overcome coordination failure in a weak-link game (Cartwright, Gillet and Van Vugt, 2013). Brands, Cooper and Fatás (2007) have also found that in a turnaround game, which is a variation of a weak-link game and models a firm suffering from low performance, leaders can increase firm’s performance by taking the initiative and increasing effort.
9
In contrast with a weak-link game, which emphasizes coordination, a public good game captures free riding under group incentives. Due to the fact that in a public good game free riding is a dominant strategy, leadership can play an essential role in increasing individual contributions. This is what is usually found in experiments. Güth et al. (2007) conducted a laboratory voluntary contribution experiment and found that leadership by example had a positive effect on contributions. Moreover, contributions were significantly higher when leaders had an opportunity to punish the followers by excluding them from the group in the next round, which shows that leaders with a possibility of punishment had more authority among the followers. Moxnes and Van der Hijden (2003) conducted a laboratory public bad experiment in which the social optimum occurred when no one contributed to the public bad, which can be interpreted as pollution. The results indicate that followers contributed significantly less when there was a leader setting a good example. Gatiso et al. (2017) conducted a field experiment in Uganda where deforestation is an important issue. The subjects had an endowment, and they were asked to decide how many trees to harvest. The authors concluded that groups with leaders were significantly more cooperative and had lower harvesting rates compared to those with no leaders.
However many researchers have found that leadership has small or negligible effect on free riding. Weber et al. (2001) found that the positive effect of leadership is often exaggerated because people falsely attribute successes or failures to leader’s ability instead of situational difficulty. In their experiment, leaders had to convey a motivational speech to their teammates. They found that subjects judged leaders as ineffective in large groups and effective in pairs when in fact cooperation occurs more often in small groups rather than in large ones (Weber et al., 2001).
Rivas and Sutter (2011) carried out an experiment where leadership was assigned exogenously or voluntary and found that exogenously appointed leaders did not affect contributions of other players. The authors concluded that while voluntary leadership increases contributions, leadership per se does not. This finding is important for the present research because it shows that the effect of leadership depends on how it is assigned and the effect is stronger when leadership is voluntary.
Overall, despite the abundance of academic literature on leadership and group incentives, it is still a question whether leadership can effectively reduce free riding and increase efficiency
10
under group incentives. Moreover, there is no clear link between leadership theories and the actual relationship between leadership styles and team performance. Hence, the purpose of this research was to investigate the effect of transformational and transactional leadership styles on inducing a socially desired action under group incentives and compare the outcome to a situation when people work individually and hence have an opportunity to choose their preferred effort.
III. Methodology
In this chapter, we discuss the features of the experimental design, the procedures that took place during the experiment and the hypotheses that were tested.
Experimental design
The experiment was designed to replicate a situation where the subjects would be rewarded based on individual effort and based on group performance. Since the experiment was going to be conducted in a secondary school, it was formulated in school terms to mitigate potential problems with understanding of the abstract setup. Participants were instructed that they had a hypothetical home assignment and 3 hours of free time to work on it. In all parts, the task was to choose how to allocate 3 hours between work and leisure. The subjects had four options to choose from: 0 hours of work and 3 hours of leisure, 1 hour of work and 2 hours of leisure and so on. The amount of time allocated to work can be interpreted in terms of effort: the more time an individual works on the assignment, the more effort he exerts on the task. Thus, further we use the terms “work time” and “effort” interchangeably.
The experiment consisted of four parts, each having one round. The first two parts were the same for all participants. The third and the fourth part differed between treatments. Hence, the experiment used within-subjects design in all parts as well as between-subject design in the third part.
The purpose of the first part was to reveal the value of leisure time and grade preferences of the participants. The subjects were instructed that the grade for the assignment would depend on the amount of time they would allocate to work: the longer they would work on the assignment, the higher their grade would be and the less free time they would have for leisure
11
activities. Hence, the trade-off involved choosing an optimal ratio of work and leisure time that would result in a preferred grade. The subjects received 1 point for giving an answer in this part. The instructions emphasized the fact that the final payoff would not depend on the choice in this part so that the students would reveal their preferences by choosing their utility-maximizing options.
The second part was a variation of a public goods game and was designed to test how people behave under group incentives. The design replicated an organizational setting where a team of employees from different departments is assigned a joint project and their bonuses depend on the project’s success. The task was to choose one of the four options of time allocation between work and leisure. In the second part the grade for the group assignment was based on the total amount of work time in a group and was the same for all group members (Table 1). It was explained that the subjects would be randomly matched with two other people in the classroom and their payoff would depend on the choices made by all three of them. Table 1. Group grade
Total work time 0 1 2 3 4 5 6 7 8 9
Grade 0 1 2 3 4 5
The payoff structure emphasized the trade-off between having more leisure time and receiving a higher grade for the group assignment:
𝜋! = 3 − 𝑡! + 𝐺( 𝑡!)
! (1)
where 𝑡! is the time allocated to work by participant 𝑖 = {1, 2, 3}, and 𝐺( !𝑡!) is the grade which depends on the sum of work hours chosen by all group members (𝑗 = {1, 2, 3}). Hence, increasing work time would decrease the amount of leisure time, however it may increase the grade for the group assignment.
In the next parts of the experiment the task and the payoff structure remained the same as in the second part: the subjects were asked to choose the number of hours that they would like to spend on a group assignment. The payoff was not revealed until the end of the experiment.
12
That way, there was no learning effect and the decisions made in previous parts did not affect future decisions.
The third part involved leadership in two forms: leading by verbal encouragement or leading by monitoring and punishment. One experimental session involved one form of leadership. The participants learned their role (leader / non-leader) at the beginning of the third part. All subjects were instructed about the procedures and the leader’s role before they made their decisions.
In the verbal encouragement treatment, the leaders were asked to convey a message to their teammates in order to encourage them to choose maximal effort. They were provided with instructions that contained a short speech that they were asked to deliver and an example of the efficient outcome for the team (Appendix. Leader’s task (verbal)). Rephrasing was allowed to make the leaders sound more convincing.
In the monitoring and punishment treatment, the leaders could observe their teammates’ choices and had an opportunity to implement punishment by decreasing the payoffs of their teammates by 2 points. The punishment was costly for the leader. Punishment of one teammate decreased leader’s payoff by 1 point. For example, in a team where two members (including the leader) chose 3 work hours and one member chose not to work on the group assignment, the payoffs were 3, 3 and 6 respectively. In case the leader decided to punish the free rider, his payoff would be 2 instead of 3 and the free rider would get 4 instead of 6. The two types of leadership selected for this thesis represent transformational and transactional leadership styles. Namely, verbal encouragement is a simplified version of what a transformational leader would do in order to motivate the followers. As was mentioned previously, a transformational leader communicates and generates awareness of the team goals. On the other hand, monitoring and punishment characterize the actions of a transactional leader who intervenes when team members deviate from standards.
The fourth part differed between the treatments. In the verbal encouragement treatment, leadership was removed in the fourth part and the participants were told that they would be assigned to teams randomly. Thus, the fourth part was intended to test whether a motivational style of leadership can have a lasting effect on effort in groups.
13
In the monitoring treatment, the fourth part involved increased punishment. The cost of punishment remained the same however the number of points that a leader could subtract from his teammates was increased from 2 to 3. Hence, the purpose of the fourth part in the monitoring treatment was to see if the power of punishment affects effort in groups.
The experimental design is summarized in the following table: Table 2. Summary of the treatments
Part 1 Part 2 Part 3 Part 4
Treatment 1. Verbal Encouragement No incentives, individual decision Group incentives, no leader Group incentives, leader, verbal encouragement Group incentives, no leader Treatment 2. Monitoring and punishment No incentives, individual decision Group incentives, no leader Group incentives, leader, monitoring and punishment Group incentives, leader, increased punishment
Public good game
The second, the third and the fourth parts of the experiment were designed to replicate a public good game in an organizational setup. In the current context, working on a group assignment is essentially contributing to the public account in the sense that marginal return from investing into the public account is lower than from investing into private account. The game has three Nash equilibria in pure strategies (see Appendix. Table 9). The first is when all three players choose 0. Then their payoff equals 3 − 0 + 0 = 3 points. Together they earn 3 ∗ 3 = 9 points. Another Nash Equilibrium occurs when everyone chooses 1. Then everyone receives a payoff of 3 − 1 + 2 = 4 points. Total surplus equals 12 points. The case when two players choose 0 and one chooses 1 is also a Nash Equilibrium as the latter is indifferent between 0 and 1.
In general, in this game choosing 0 is a weakly dominant strategy. Table 3 describes individual payoffs depending on the individual choice and the choice of two other players together. For instance, in case each of the two other players chooses 3 hours, a third player is indifferent between choosing 0 or 1. However if two other players choose 2 and 3 hours respectively, the third player is better off choosing 0 work hours.
14
The efficient outcome occurs when all team members choose 3 hours. Then individual payoff equals 3 − 3 + 5 = 5 points and total surplus is 15 points. However it is not an equilibrium in this game because the players have an incentive to deviate and choose 0 to get a payoff of 6. For that reason, leadership can play an important role in achieving a socially desirable outcome.
The game was the same in the second, third and the fourth part of the experiment, hence we could compare which leadership style is more likely to increase effort under group incentives. Leadership did not change equilibria in the game since in both treatments choosing zero effort remained a weakly dominant strategy. In the verbal encouragement treatment, there were no monetary consequences and the leader did not observe the chosen effort, so it was always beneficial to choose 0. In the monitoring treatment, the leader’s payoff decreased with punishment, so non-leaders could anticipate that the leaders would never implement costly punishment, and thus selecting 0 effort was their best option. Nevertheless, the efficient outcome was the same across the treatments and it occurred when all team members chose the most costly option.
Table 3. Individual payoff in a public good game
Total effort of two other teammates
Individual effort 0 1 2 3 4 5 6
0 3 4 4 5 5 6 6
1 3 3 4 4 5 5 6
2 2 3 3 4 4 5 5
3 2 2 3 3 4 4 5
Note: the bold numbers represent the maximum payoffs for a given amount of effort of the other players.
Procedures
The experiment was run in seven sessions: four for the verbal encouragement treatment and three for the monitoring and punishment treatment. Each session contained 9 to 24 participants. A total of 139 subjects participated in the experiment. The subjects were
15
students from two secondary schools in Moscow. The experiment was run on paper in the classrooms. Each session consisted of four parts.
Upon entering the classroom, all subjects randomly picked a card with a participant ID number. They were asked to take a seat with the same number. On their tables, the participants had a decision sheet for part 1. First, they were asked to fill in information about their age, gender and average grade on the top of the decision sheet for part 1 (Appendix. Decision Sheet Part 1). After everyone was finished, the instructions for part 1 began. The participants were instructed that they should choose one of the options of time allocation which would be the most suitable for them. It was emphasized that the answers were anonymous and that the payment in this part would not depend on the chosen option to avoid socially desirable answers. The subjects were given 1 minute to make a decision. After that, the decision sheets were collected.
Then the beginning of the second part was announced. The participants received new decision sheets (Appendix. Decision Sheet Part 2) and listened to the instructions. The table with a group grade and the payoff formula were put on the blackboard. To ensure that everyone understood the terms of the experiment, we asked the participants several control questions that were discussed altogether and put on the blackboard as examples. The subjects were instructed that their payoff would be calculated at the end of the experiment based on random matching into groups. The decision sheets specified four options of time allocation. The subjects had 1 minute to choose their preferred option. The decision sheets were then collected.
The procedures in the third part of the experiment were different in two treatments. In the verbal encouragement treatment, first, everyone received the decision sheets that indicated the role of participants (leader / non-leader) (Appendix. Decision Sheet Part 3) and the leaders received their instructions with a short speech (Appendix. Leader’s Task (a) verbal). Figure 1 shows team composition in a classroom and how leadership was assigned. After the leaders finished reading the instructions, the teams were asked to gather around their leaders who had one minute to convey a message written in the instructions. After one minute, the participants were asked to take their seats and make a decision privately. The task was identical to that in the second part. The decision sheets and leaders’ instructions were collected after all participants had made their decisions.
16
Figure 1. Team composition. Verbal encouragement treatment
Row 1 Row 2 Row 3
Team 1 non-leader Team 1 leader Team 1 non-leader Team 6 non-leader Team 6 leader Team 6 non-leader Team 2 non-leader Team 2 leader Team 2 non-leader Team 7 non-leader Team 7 leader Team 7 non-leader Team 3 non-leader Team 3 leader Team 3 non-leader Team 8 non-leader Team 8 leader Team 8 non-leader Team 4 non-leader Team 4 leader Team 4 non-leader Team 9 non-leader Team 9 leader Team 9 non-leader Team 5 non-leader Team 5 leader Team 5 non-leader Team 10 non-leader Team 10 leader Team 10 non-leader
Team composition in the monitoring and punishment treatment is depicted in Figure 2. The difference with the verbal encouragement treatment was that the leaders were separated from the teammates so that they would not base their punishment on the personalities of their teammates. At the beginning of the third part, the subjects received the decision sheets that indicated their role (leader / non-leader) (Appendix. Decision Sheet Part 3). The participants were told about the leader’s task but they were not told who their leader would be. Then the subjects were asked to choose one of the four options on their decision sheets as in the previous parts. When everyone was ready, the decision sheets of the leaders were collected and the decision sheets of non-leaders were given to their team leaders. With that, the leaders received additional decision sheets for punishment (Appendix Leader’s Task (b) monitoring). They could indicate whether they wanted to decrease their teammates’ payoffs by two points. It was clarified that if a leader decided to decrease his teammate’s payoff, then the leader would lose one point. After the leaders made their choice, all decision sheets were collected.
17
Figure 2. Team composition. Monitoring and punishment treatment
Row 1 Row 2 Row 3
Team 1 non-leader Team 6 leader Team 1 non-leader Team 6 non-leader Team 1 leader Team 6 non-leader Team 2 non-leader Team 7 leader Team 2 non-leader Team 7 non-leader Team 2 leader Team 7 non-leader Team 3 non-leader Team 8 leader Team 3 non-leader Team 8 non-leader Team 3 leader Team 8 non-leader Team 4 non-leader Team 9 leader Team 4 non-leader Team 9 non-leader Team 4 leader Team 9 non-leader Team 5 non-leader Team 10 leader Team 5 non-leader Team 10 non-leader Team 5 leader Team 10 non-leader
The fourth part in the verbal encouragement treatment replicated part 2 of the experiment. The subjects were instructed that they would be matched into teams randomly and they were asked to make a decision about their preferred work time on a group assignment. In the monitoring treatment, the procedures in the fourth part were the same as in part 3. In addition, the instructions emphasized the fact that the leaders would have an opportunity to decrease their teammates’ payoffs by 3 instead of 2 points.
Hypotheses
The first purpose of this research was to test if people choose their preferred allocation of work and leisure time under group incentives or if they start free riding in a group. We expected to see that subjects would pick lower effort in the second part of the experiment compared to the first part. Assuming that the subjects act rationally and select the option that maximizes their utility in the absence of monetary incentives, it was reasonable to expect that they would decrease effort in the second part because they would not receive full proceeds from effort increases (Baker, Jensen and Murphy, 1988). Taking into account the fact that free riding rarely occurs in one-shot experiments (Andreoni, 1988; Fischbacher and Gachter, 2010), the reduction of effort under group incentives was underestimated in our experiment.
18
H1: Effort under group incentives is likely to be lower than the preferred level chosen in the absence of incentives.
The second hypothesis concerns the effect of leadership on individual performance under group incentives. Several studies have shown that leadership can help achieve a socially desirable outcome and increase cooperation in experiments (Moxnes and Van der Hijden, 2003; Brands, Cooper and Fatas, 2007; Guth et al., 2007; Cartwright, Gillet and Van Vugt, 2013; Gatiso et al., 2017). Multiple studies conclude that transformational leadership is positively associated with performance (Schaumbroeck, Lam and Cha, 2007). Thus, we expected to see that effort would increase in the third part in the verbal encouragement treatment as compared to the second part.
H2.a: Leading by verbal encouragement increases effort under group incentives.
Likewise, we expected to see that effort would increase in the monitoring and punishment treatment. Fehr and Gächter (2002) found that altruistic punishment of defectors can explain cooperation in a public good game. It has been shown that in the presence of punishment opportunities there is less free riding (Fehr and Gächter, 2000). Similarly to that, Guth et al. (2007) found that contributions were higher when a leader was granted exclusion power. We hypothesized that leading by monitoring and punishment can create an incentive to choose maximum effort in order to minimize the risk of being punished for shirking.
H2.b: Leading by monitoring and punishment increases effort under group incentives.
One of the purposes of using two variations of leadership was to compare the effectiveness of different leadership styles. Existing literature leads to a conclusion that verbal encouragement as a leadership style is more likely to lead to high effort level in a group as opposed to monitoring and punishment (Deci and Cascio, 1972; Bass, 1990). Verbal encouragement involves no economic consequences for the team members. Thus, the leader’s success in reducing free riding depends on the leader’s ability to convince the teammates to cooperate. Monitoring and punishment, on the other hand, imposes a risk of losing points in case a leader decides to impose punishment, which is costly for the leader. Therefore leader’s success depends on whether his teammates consider punishment a credible threat. More importantly, the effect of leadership on effort depends on what type of behavior non-leaders expect to be punished for. Hence, if they expect the leader to punish free riding, they will
19
exert maximum effort. Comparing effort levels between two leadership variations allows making conclusions about which leader strategy is more likely to motivate group members. H3: Verbal encouragement is more likely to induce higher effort than monitoring and punishment.
The fourth part of the experiment with leadership in the form of verbal encouragement replicated the second part of the experiment with group incentives and stranger design. We expected to see that the effort level would decrease compared to the part with leadership. One of the reasons may be that in the absence of a leader the group members would no longer feel motivated to exert high effort. However, another reason may be that communication was no longer possible in the fourth part. It has been shown that face-to-face group communication results in a reduction of free riding behavior, however group optimality decreases in periods with no communication (Isaac and Walker, 1988). Secondly, research indicates that the absence of face-to-face contact between group members leads to lower contributions compared to a situation when group members meet each other (Bochet, Page and Putterman, 2006).
H4: The positive effect of leadership on effort under group incentives vanishes when leadership is removed.
The fifth hypothesis is related to the effect of increased punishment on effort. Fehr and Gächter (2000) have found that cooperators willingly punished free riders despite the fact that punishment was costly, and punishment was higher for lower contributions. Taking into account the fact that the cost of punishment for leaders remained the same in the fourth part, it is possible that they would be more willing to punish free riders in the increased punishment treatment. Hence, exerting higher effort would make it less likely for non-leaders to be punished in the fourth part. Because of that, we expected to see that effort would increase in the fourth part in the monitoring treatment.
20
IV. Results
In this section, we present the results of the experiment and test the hypotheses. This section also contains the analysis of leaders’ behavior and conclusions regarding the effect of leadership on free riding in the experiment.
Descriptive statistics
A total of 139 students from two secondary schools in Moscow took part in the experiment. 87 students participated in 4 sessions with leadership in the form of verbal encouragement and 52 students participated in 3 sessions with leadership in the form of monitoring and punishment. Participants were randomly assigned to treatments. Due to the fact that in some sessions the number of participants was not a multiple of three, 7 observations had to be dropped because the teams were not complete in the third and the fourth parts of the experiment. The final sample consists of 132 observations.
In the first part of the experiment, all participants indicated their gender, age and their average grade on a five-point scale. Summary statistics are presented in Table 4. The proportion of male and female leaders, their average age and average grade are indicated in brackets.
Table 4. Summary statistics Number of participants
Gender
Male Female Average age Average grade
Verbal
encouragement 81 38% (41%) 62% (59%) 15.94 (15.93) 4.42 (4.33)
Monitoring and
punishment 51 51% (59%) 49% (41%) 15.65 (15.53) 4.09 (4.12)
Note. The proportion of male and female leaders, their average age and grade are in brackets. By visual inspection, the average grade in the verbal encouragement treatment was somewhat higher than in the monitoring and punishment treatment. According to the Fisher’s exact test, there is a statistically significant difference in the proportions of people who reported ‘3’, ‘4’ or ‘5’ as their most common grade between treatments (p = .000). There is no statistically significant difference in proportions of male and female students between treatments (p =
21
.206). There is a marginally significant difference in the distribution of ages between treatment groups (p = .099).
The first finding is that in the first three parts in both treatments the majority of subjects chose high levels of effort (Figure 3). This is consistent with the previous research on the public good games (Andreoni, 1988; Brands, Cooper and Fatás, 2007). There may be several reasons for a small number of low contributions. Firstly, since the experiment was carried out in a school setting, the subjects might have given socially desirable answers even though they were instructed that their responses would be completely anonymous. It could especially be the case in the first part where there were no monetary incentives and the decision was made in a hypothetical setting, so choosing a socially desirable option was costless. Under group incentives in the second and the third part choosing low work time could also be associated with negative feelings such as fear of social punishment, for example in the form of exclusion from a group.
Hypotheses testing
Figure 4 presents the average levels of effort chosen by the subjects in the experiment, divided by treatment and all together in parts 1 and 2.
1% 7% 2% 8% 0% 4% 4% 14% 63% 45% 10% 45% 30% 45% 86% 37% Part 1. Individual incentives Part 2. Group incentives Part 3. Verbal Encouragement
Part 3. Monitoring and punishment % of s u b je cts Individual effort Figure 3.
Individual effort choices
22
The first hypothesis states that people choose lower effort under group incentives compared to their preferred effort selected in the absence of incentives. Contrary to what was expected, the difference between effort choices in the first and the second part was negative (-.121) and significant according to the Wilcoxon rank-sum test (N = 132, z = -1.818, p-value = .069). Table 5 summarizes how the subjects reacted to the introduction of group incentives (Part 2) and leadership (Parts 3, 4). The numbers in the table represent the proportion of people who increased, did not change or decreased their effort between two subsequent parts. It can be seen that the majority of the subjects (59.9%) did not change their choice in the second part and only 15.1% of participants decreased their effort under group incentives compared to a situation without incentives. This means that 25% of subjects preferred to work more when their grade was dependent on the total effort exerted by all three team members compared to a situation with no incentives where they could choose their preferred work time and grade. Hence, we reject the first hypothesis.
2,212 2,333 2,247 2,407 2,827 2,519 2,157 2,216 2,157 2,314 2 2,2 2,4 2,6 2,8 3 Part 1.Individual incentives Part 2. Group incentives Part 3. Group incentives + leadership Part 4 A ve rage e ffor t Figure 4. Average effort choices
23
Table 5. Summary reactions to treatments
Part 2. Part 3. Part 4.
Group incentives Verbal Monitoring No leader,
verbal Increased punishment Increase 25% 42% 15.7% 7.4% 17.6% No change 59.9% 51.8% 66.7% 64.2% 70.6% Decrease 15.1% 6.2% 17.6% 28.4% 11.8% N 132 81 51 81 51
Note: the numbers represent the proportion of people that increased, did not change or decreased their effort between the subsequent parts of the experiment.
The second hypothesis is related to the effect of leadership on effort under group incentives. According to the Wilcoxon rank-sum test, there is a statistically significant difference between effort levels in part 2 and part 3 of the experiment in the verbal encouragement treatment (N = 81, z = -4.593, p-value = .000). From Table 5 it can be seen that in the third part in the verbal encouragement treatment 42% of subjects increased their effort compared to part 2 and only 6.2% chose lower effort than in part 2.
We further analyzed how the subjects with different effort levels in the second part reacted to leadership in the third part in order to draw a conclusion about what types of participants were affected by leadership the most. Table 6 summarizes the choices of subjects in the second part of the experiment and shows how they responded to leadership in the third part. As can be seen, 80% of those who chose 2 hours in the second part increased their effort in the part with leadership in the form of verbal encouragement. Moreover, 92% of those who already exerted maximum effort under group incentives did not change their choice in the third part in the verbal encouragement treatment. This gives a reason to believe that leadership in the form of verbal encouragement motivated people to exert maximum effort under group incentives. The fact that 100% of people who chose 0 or 1 hour under group incentives increased their effort in the third part should be however interpreted with caution due to a small number of observations.
In the monitoring treatment there was no significant difference between efforts in part 2 and part 3 (N = 51, z = .287, p-value = .774). The majority of subjects (66.7%) chose the same
24
effort as in the part with no leadership and only 15.7% increased their effort (Table 5). Moreover, 31.6% of those who chose maximum effort in part 2 chose lower effort in the third part with leadership in the form of monitoring and punishment (Table 6). Hence, we can conclude that this type of leadership does not increase effort under group incentives.
Table 6. Summary reactions to leadership
Verbal encouragement Monitoring
Effort
Part 2 Increase
No
change Decrease N Increase
No change Decrease N 0 100% 0% - 1 0% 100% - 1 1 100% 0% 0% 5 50% 50% 0% 6 2 80% 14.3% 5.7% 35 20% 68% 12% 25 3 - 92.5% 7.5% 40 - 68.4% 31.6% 19 overall 42% 51.8% 6.2% 81 15.7% 66.7% 17.6% 51
Note: the numbers represent the proportion of subjects that increased, did not change or decreased their effort in response to leadership divided by the effort levels in the second part. In addition to that, we tested how effort in the two treatments compared to that chosen under individual incentives. Wilcoxon rank-sum test indicates that the difference between efforts in part 1 and part 3 is significant in the verbal encouragement treatment (N = 81, z = -5.891, p-value = .000). On average, effort chosen in the part with leadership in the form of verbal encouragement was 21.4% higher than under individual incentives. There was however no significant difference between effort choices under individual incentives and group incentives with leadership in the monitoring and encouragement treatment (N = 51, z = -.226, p-value = .821). This suggests that leadership in the form of verbal encouragement can enhance performance as compared to a situation when an individual chooses his preferred option costlessly, while a monitoring style of leadership does not affect effort choices.
The third hypothesis states that verbal encouragement is more likely to induce higher effort than monitoring and punishment. This was supported by the results: on average, the subjects chose higher effort in the third part with leadership in the form of verbal encouragement compared to monitoring and punishment (Figure 4). Two-sample Wilcoxon rank-sum test indicates that the mean difference between the effort choices in the monitoring and verbal
25
encouragement treatments is statistically significant (N = 132, z = -5.780, p-value = .000). However, comparing means was not sufficient due to the fact that we found that there was a statistically significant difference in the average grades of participants between treatments. Therefore, we ran an ordered probit regression, which also allowed controlling for behavior in previous parts. The dependent variable was effort in the part with leadership. For the first model, we only included the binary variable which equaled 1 for the verbal encouragement treatment and 0 for the monitoring and punishment treatment. The second model included effort choices in the first two parts. The third model included the control variables. The coefficient of the treatment variable (Verbal) was positive and significant in all three models (Table 7). Analysis of the marginal effects also indicated that in the verbal encouragement treatment the subjects were more likely to choose the highest effort and less likely to choose effort below 3 (Appendix. Table 10). This holds for all models that we used in this part. This leads to a conclusion that verbal encouragement as a leadership style is more likely to induce higher effort in a group compared to monitoring and punishment.
Table 7. Comparison of effort between treatments
Variable (1) (2) (3) Verbal 1.305*** (.234) 1.291*** (.242) 1.249*** (.263) T1 – .321 (.226) .221 (.236) T2 – .467*** (.178) .492*** (.182) Controls No No Yes Obs 132 132 132 Prob chi2 > 0 .000 .000 .000 Pseudo R2 .144 .202 .219
Note. Dependent variable: individual effort in Part 3. The table presents coefficients from the ordered probit regressions. Verbal variable equals 1 if leadership was in the form of verbal encouragement and 0 otherwise. T1 is the effort choice in part 1, T2 is the effort is part 2. Controls include age, gender and average grade. Standard errors are in parentheses. ***p< .01.
The fourth and the fifth hypotheses required comparing effort levels between the third and the fourth parts of the experiment within the treatments. We expected to see that the effort would drop in the verbal encouragement treatment and increase in the monitoring and
26
punishment treatment. On average, subjects chose 7.4% lower effort in the fourth part compared to the third part in the verbal encouragement treatment. The difference was significant according to the Wilcoxon signed-rank test (N = 81, z = 3.244, p-value = .001). At the same time, there was no significant difference between effort in part 2 and part 4 in the verbal encouragement treatment (N = 81, z = -1.454, p = .146). Hence, leadership in the form of verbal encouragement did not have a lasting effect on the subjects since removing a leader in the verbal encouragement treatment resulted in a decrease of effort.
In the monitoring treatment the increased punishment on average led to a higher level of effort, however the increase was statistically insignificant (N = 51, z = -.913, p = .361). Combined with the previous finding that leadership in the form of monitoring and punishment had no statistically significant effect on effort under group incentives, this result suggests that increased punishment did not create stronger incentives for the non-leaders to exert higher effort. Due to the fact that punishment was costly for the leaders, the team members may have expected that the leaders would not punish low effort. Further we discuss leaders’ punishing behavior in more detail.
Leaders’ behavior
The subjects learned their role (leader or non-leader) at the beginning of the third part of the experiment. Average effort choices of leaders and non-leaders are shown in Figure 5. According to the two-sample Wilcoxon rank-sum test, there is no statistically significant difference in effort choices between leaders and non-leaders in the third part of the experiment in the verbal encouragement treatment (N = 81, z = -.531, p-value = .596) as well as in the monitoring and punishment treatment (N = 51, z = 1.092, p-value = .275). Moreover, there is no difference in how leaders and non-leaders changed their choice in response to leadership in the verbal encouragement treatment (N = 81, z = .130, p-value = .897) and in the monitoring and punishment treatment (N = 51, z = .336, p-value = .737). Hence, we did not find evidence that leaders’ behavior was significantly different from the non-leaders’ behavior in the experiment.
27
The punishing behavior of the leaders was in line with the previous research. Firstly, punishment occurred quite rarely. Out of seventeen leaders, four punished their teammates in the third part (23.5%) and four punished their teammates in the fourth part (23.5%). In both parts 50% of leaders who punished were male and 50% were female. Only one leader punished his teammates in the third and the fourth parts and only one leader punished both teammates, the rest punished only one.
The choices that the participants have been punished for are shown in Table 8. Overall, very few subjects got punished for choosing low effort. On the other hand, many leaders punished high effort. It may explain why leadership in the form of monitoring and punishment had insignificant effect on effort. Non-leaders may have correctly anticipated that the leaders would not punish specific behavior such as free riding and hence did not change their choice in the parts with leadership.
As can be seen from the Table 8, the leaders varied in the type of behavior that they punished. Kosfeld and Rustagi (2015) distinguish between different leadership styles: leaders that do not punish, leaders that punish those who contributed less than others, leaders that punish unequal contributions and antisocial leaders who punish those who contributed the full endowment. In our experiment 59% of leaders never punished, 17% punished those who
2,2 2,5 2,9 2,3 2,4 2,8 2,6 2,1 2,0 2,4 2,1 2,2 2,2 2,3 1,5 2,0 2,5 3,0
Part 1 Part 2 Part 3 Part 4
A ve ra ge e ff or t Figure 5. Average effort choices of leaders and non-leaders Leaders, verbal encouragement Non-leaders, verbal encouragement Leaders, monitoring and punishment Non- leaders, monitoring and punishment
28
chose effort below 3, 12% punished those who contributed the full endowment and 12% punished unequal contributions.
Table 8. Summary punishments
Part 3 Part 4 Effort choice Number of participants (non-leaders) Number of punishments Number of participants (non-leaders) Number of punishments 0 1 0 1 1 1 5 2 5 1 2 13 0 12 1 3 15 3 16 1 overall 34 5 34 4
Note: the table describes how many non-followers chose different levels of effort in the parts with leadership and how many of them were punished for that choice.
Leadership and free riding
One of the goals of this research was to test if leadership could reduce free riding in groups. To test that, we divided the subjects into three categories according to their reaction to group incentives and analyzed the reactions to leadership within the categories.
The first category was the free riders. In general, free riding can be defined as exerting the effort below the preferred level in order to maximize individual payoff. Hence, we defined free riders as those who decided to work less hard under group incentives than they did in the first part without incentives. The second category included those who have chosen their preferred effort under group incentives, in other words, the subjects that were not affected by the change in incentives. The third category referred to the subjects who have chosen higher effort in the second part than in the first one. Hence, those participants decided to work harder in a group than on their own. The reactions of subjects to leadership in part 3 are shown in Figures 6a and 6b.
Around 15% of subjects acted as free riders in the second part in both treatments (14.8% in verbal encouragement and 15.7% in monitoring treatment). However, while in the verbal encouragement treatment 91.7% of free riders increased effort with leadership, only 37.5% of
29
free riders in the monitoring and punishment treatment responded to leadership with increased effort. This suggests that leadership can reduce free riding, however the effect may be stronger if a leader encourages his teammates instead of threatening them with punishment.
The majority of subjects in both treatments chose their preferred effort under group incentives: 59.3% in the verbal encouragement treatment and 60.8% in the monitoring and punishment treatment. Similarly to the previous category, the subjects were incentivized to choose higher effort more often in the verbal encouragement treatment than in the monitoring and punishment treatment. Moreover, almost 20% of subjects in this category have decreased their effort in response to leadership in the form of monitoring and punishment. Contrary to that, 45.8% of subjects in this category increased their effort as a response to leadership in the form of verbal encouragement. This can be attributed to the fact that the threat of punishment erodes intrinsic motivation (Deci and Cascio, 1972) and positive verbal reinforcement increases intrinsic motivation (Deci, 1972). Thus, those who did not decrease their effort under group incentives and hence were intrinsically motivated to work hard might have been demotivated by the threat of leadership. On the contrary, verbal encouragement had a motivating effect on almost half of this category in the verbal encouragement treatment. The next category included subjects that increased their effort under group incentives compared to their individual choice in the first part. In contrast to theoretical predictions, this was quite a large fraction of participants, namely 25.9% of participants in the verbal encouragement treatment and 23.5% of participants in the monitoring and punishment treatment chose higher effort in the second part compared to the first part. Most of these subjects selected maximum effort in part 2 (90.5% in the verbal encouragement treatment and 100% in monitoring and punishment treatment), hence they could not further increase effort in the third part. However, the fraction of subjects that decreased their effort in the part with leadership was higher in the monitoring and punishment treatment (25%) than in verbal encouragement treatment (14.3%), which suggests that monitoring style of leadership was less effective in preventing free riding among those who worked harder in group than individually compared to the encouragement style of leadership.
30
V. Discussion
This section is dedicated to the discussion of the results, limitations of the current research and suggestions for further research.
0,0% 4,2% 14,3% 8,3% 50,0% 81,0% 91,7% 45,8% 4,8%
Reaction Group < 0 (14,8%) Reaction Group = 0 (59,3%) Reaction Group > 0 (25,9%)
Figure 6a. Reaction to group incentives and leadership Verbal encouragement treatment
Reaction Leaderhsip < 0 Reaction Leadership = 0 Reaction Leadership > 0
0,0% 19,4% 25,0% 62,5% 64,5% 75,0% 37,5% 16,1% 0,0%
Reaction Group < 0 (15,7%) Reaction Group = 0 (60,8%) Reaction Group > 0 (23,5%)
Figure 6b. Reaction to group incentives and leaderhsip Monitoring and punishment treatment
31
Discussion
One of the goals of the present study was to compare the effectiveness of group and individual incentives in inducing higher individual effort. The results suggest that people often increase their effort when they are put into groups rather than when they are working on their own. One of the reasons why people decided to work harder in the second part of the experiment may be that being a part of a group was associated with cooperative values (Qin, Johnson and Johndon, 1995) such as a feeling of responsibility or unwillingness to let down the teammates. However another explanation could be that cooperative behavior was affected by the cultural background. Herrmann and Thöni (2008) replicated a study on cooperation that was initially carried out in Switzerland with undergraduate students in Russia, and found that the share of free riders was significantly lower than in the Swiss subject pool. It is possible that cooperative behavior is characteristic of Russian subjects, hence they were inclined to choose high effort under group incentives. Besides that, the subjects may have been reluctant to choose low effort under group incentives because they were not familiar with the free rider strategy due to little experience with group assignments. In Russian secondary schools home assignments rarely involve group work, hence it might have been difficult for the subjects to imagine what they would do and they opted for high effort.
Furthermore, the results provide evidence for the superiority of transformational over transactional leadership: while verbal encouragement had a significant positive effect on effort in groups, monitoring and punishment did not incentivize the subjects to exert more effort. It is possible that leaders managed to persuade the teammates to cooperate and choose high effort to increase the total surplus in the group. However, this may have resulted from the fact that verbal encouragement treatment involved communication while the monitoring treatment did not. As was mentioned before, face-to-face interaction increases contributions in a public good game (Bochet, Page and Putterman, 2006), hence cooperation may have resulted from the fact that team members met each other before making a decision in the verbal encouragement treatment.
Besides that, this result can be attributed to small size of groups (Rivas and Sutter, 2011). It has been previously shown that cooperation is achieved more easily in small groups
32
compared to large ones, so in our case, increased effort may have resulted from the fact that the leaders had to persuade only two other persons.
The fact that leaders did not choose lower effort in the third part in the verbal encouragement treatment may imply that leaders did not perceive their role as an opportunity to make others work harder and hence have more free time. Probably, being a leader was interpreted as a responsibility more than a privilege.
Limitations
The limitations of the current research concern the design and the execution of the experiment. Firstly, the number of participants in the monitoring treatment was quite small, which may be the reason why the effect of leadership on effort was statistically insignificant in the treatment with punishment. Even though we conducted the experiment at two secondary schools, the number of potential participants was restricted due to school limitations.
Since one of the purposes of this thesis was to expand the scope of research on the effect of leadership to see how people behave as compared to a situation when they make their decision individually and choose their preferred option, the experiment was designed in a way that the task would be similar in all parts. However, it resulted in another limitation of the analysis because effort in the first part and in parts with group incentives may not be entirely comparable since in the first part there were no incentives. Nevertheless, even though the measure of effort was not perfect, the results were consistent with the predictions and current scholarly papers on group incentives.
Another limitation comes from the fact that the pupils who took part in the experiment might have faced problems understanding the experiment. Some sessions were carried out in English as they were conducted during the English lessons. Hence, even though the terms of the experiment and the payoff structure were thoroughly explained and several examples were given, it is possible that there remained a certain level of confusion among the participants. Moreover, for the secondary school pupils, it might have been difficult to imagine that they had to work on a group project, as it is not a common practice in the Russian school system.
33
In addition, it is possible that the subjects gave the answers that they thought were expected from them, maximum work hours in particular. For all participants, this was their first experience with economic experiments. They might have not trusted the experimenters not to disclose the answers to their teachers or parents. However, it was repeatedly noted throughout the experiment that the answers would remain confidential and anonymous. Thus we do not expect that this could bias the results significantly, especially taking into account the fact that some subjects chose low effort even in the first part where choosing maximum effort was costless.
As was mentioned previously, there was a significant difference in the average grades of the subjects between treatments. Namely, the participants in the verbal encouragement treatment reported on average higher grades than in the monitoring treatment. The average grade can be interpreted as a proxy for cognitive ability. Hence, even though the assignment of participants to the treatments was random, the group composition might have driven the results to a certain extent. It is possible that the subjects managed to understand the experimental design better in sessions with verbal encouragement treatment. In that case, insignificant results in the monitoring treatment could be attributed to poor understanding of the experiment. However, due to the fact that we controlled for the average grade while comparing the effect of leadership between treatments, we do not expect that the differences in the subject pools affected the conclusions.
Further research
As was mentioned, one of the limitations of the current study is that the subjects were asked to imagine a situation and make a decision in a hypothetical setup. For the further research, it could be interesting to see if using a real-effort task would change the results. We suppose that the hypothetical setup underestimates the effect of leadership because the subjects may be prone to give the same answers in all parts.
Furthermore, it could be useful to replicate this study with communication in both treatments to disentangle the effects of personal interaction and leadership. Alternatively, this study could be conducted with larger teams to test if the team size mitigated the positive effect of leadership.
34
VI. Conclusion
Leadership can be thought of as a practical way of aligning interests within a group, overcoming coordination failures and reducing free riding. In the past, many scholars have researched the effect of leadership on coordination and cooperation in experiments. A distinguishing feature of this thesis is that we compared the effect of two leadership styles, namely encouragement and punishment, on individual choices under group incentives. The scope was also extended to compare the outcomes under group incentives with leadership and in a situation where subjects make individual decisions with no incentives.
The experiment was designed to replicate a situation where employees are working on a group project to test if leadership can increase individual effort in the presence of an opportunity to free ride. The results indicated that encouragement was more likely to have a positive effect on effort under group incentives compared to leadership in the form of monitoring and punishment. This result should however be interpreted with caution because the positive effect of verbal encouragement may be partially attributed to face-to-face communication within teams. Furthermore, the effect of encouragement did not last when participants were matched into new teams and there was no leader in the group.
The results also suggest that the monitoring style of leadership that included punishment did not increase individual effort choices compared to a situation when the subjects choose their preferred options with no incentives. This may be attributed to the fact that team members correctly anticipated that leaders would vary in the type of behaviour they would punish for, hence increasing effort would not necessarily decrease the probability of being punished. Moreover, some people may have been demotivated by the threat of punishment as it decreases intrinsic motivation.
Overall, this thesis provides a useful insight for organizations and institutions and can be used in a way to increase performance in group tasks.
