- Master Thesis -
The Gift Exchange Game:
How Robust is Reciprocal Behaviour in the Public Sector?
Name: Noven Kusuma Nugraheni Student Number: 11386770 Master: Economics and Business
Track: Behavioural Economics and Game Theory Supervisor: Dr. Joël van der Weele
1
Statement of Originality
This document is written by Student Noven Kusuma Nugraheni who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
2
Abstract
This research investigates to what extent the features of the public sector such as separation of ownership and control as well as public wages scheme influence worker’s reciprocal behaviour. In particular, it compares the standard trilateral gift exchange game with a single owner-manager which employs two workers with the gift exchange game where the institution is owned by the government but controlled by the manager. Although the results could not be statistically confirmed, all the findings indicate that there is a positive wage-effort relationship both in absolute wage and relative wage comparison, even when the manager does not share the profits. This suggests that workers may feel reciprocal towards the one who pays for the wages. In the treatment where wages are exogenously given, the results suggest that workers also feel reciprocal towards the institution as a whole, regardless of its composition.
I. Introduction
Reciprocity is a critical factor for understanding human pro-social behaviour in economic transactions. It involves the interaction of giving and returning between economic agents. In the context of labour interactions, reciprocity has received considerable theoretical attention as a strong motivator underlying workers’ behaviour (Rabin, 1993; Fehr et al., 1998; Dufwenberg and Kirchsteiger, 2004; Sobel, 2005; Fehr and Fischbacher, 2006). Typically, workers tend to exert high effort in response to the high wage offered by the employer. However, will a higher wage offer always lead to more effort? An increase in wage may have considerable positive effects on the effort in one company, but much less in another. The kind of business, organization structure, and the work environment could cause this difference.
Consider the following situation. Imagine you work in the public institution and are mandated by the manager to perform a task with your co-worker. You both do an excellent job for the organization. However, when it comes to the payment period, you learn that your wage is lower than the wage of your co-worker because you have fewer work experience and lower
3
grade level. How do you feel? Would you still reciprocate your manager by giving your best effort knowing that you get a lower pay than your co-worker? And what if your manager who sets the wages is not the one who pays for the wages? How the features of the public sector environment influence the worker’s reciprocal behaviour is examined in this paper.
The jobs in the public sector have special characteristics that differ from those in private organizations. First, the jobs in the public sector are classified by a graded scale (Osborne and Gaebler, 1992). In most countries, the pay scale in the public sector is much more rigid and deterministic than in the private sector in which seniority and education tend to be rewarded independently of performance (Osborne and Gaebler, 1992; Postel-Vinay, 2015). This system makes it difficult for a manager to fire a poor worker or reward a good one with a higher pay (Osborne and Gaebler, 1992).
Secondly, the basic salary is publicly and centrally regulated by the government, in the sense that every worker has a possibility to observe his/her own salary and the salary of others according to the grade they belong.1 Although it is considered a transparency, this public wage setting may arise negative effects when the workers compare their wages with the wages of their co-workers. Discovering the existence of pay differentials between oneself and co-workers may urge perceptions of unfairness, especially when they do a comparable work (Nosenzo, 2013). This may ultimately lead to reduced job satisfaction and lower effort provided by the workers.
Another notable feature is the multi-level hierarchy, in which there is a separation between the ownership and control. This implies that the government who owns the public institution is not the one who directly controls the management and makes executive decisions. They typically delegate the management authority to the managers, where part of the managers’ job is to determine wages or incentives. For instance, most public sector institutions such as in Australia, Switzerland, and United States nowadays have implemented performance-based pay systems in which the workers also receive additional incentives according to their
1For example, the basic salary for all public servants in Indonesia is stated in the Government Regulation No.
4
performance evaluation aside from the basic salary (Ketelaar et al., 2007). Although the manager sets the wage for the workers, he/she is not a residual claimant. It implies that the manager does not bear the full wage cost and thus the manager does not receive the full benefit generated from the worker’s reciprocal behaviour (Maximiano et al., 2013). This separation of ownership and control may diminish the reciprocal behaviour from the workers because exerting higher effort will not benefit the manager who sets the wage.
This research aims at investigating these issues by conducting an experiment on the gift exchange game. Specifically, the research question addresses how reciprocity works (the robustness of gift exchange game) in the public sector environments where the ownership and control are separated and wages are publicly known. In its most common setting, the game consists of two stages. At the first stage, the employer determines the wage transfer to the worker. After observing the wage that he/she will get, the worker then decides how much effort to supply. Effort is costly to the worker but increases employer’s profit. Although the standard economics assumption predicts that individuals are selfish, many experimental researches find a significantly positive correlation between wage and effort (Fehr et al., 1993; Fehr et al., 1998; Brandts and Charness, 2004; Charness, 2004; and Falk, 2007), indicating the existence of reciprocal behaviour.
The experiment in this paper considers three treatments, the first of which corresponds to the standard trilateral gift exchange game and serves as the baseline. In this treatment, the ownership and control are assigned to a single person i.e. the owner-manager to determine wages for both workers. Treatment 1 considers a situation in which the institution is owned by the government but controlled by the manager. The ownership and control are fully separated so that the government is a full residual claimant of the institution’s earnings (Maximiano et al., 2013). Finally, Treatment 2 considers an external party (i.e. wages were randomly chosen by the experimenter) to decide on the actual wages. This treatment represents the feature of public sector environment in which the wage is determined exogenously by a regulation. In this treatment, the member of the institution has no control over the wages which can be meaningful to observe whether control per se is important for the wage-effort relationship.
5
The main finding suggests that the wage-effort relationship does not significantly differ across treatments. The workers still reward the high wage offered with high effort and punish the low wage offered with low effort even when the ownership and control are separated. The results support the hypothesis that the workers are reciprocal towards the one who pays the wage. Given that in the treatment where wages are determined exogenously the proportion of reciprocal behaviour is smaller, the hypothesis that workers feel reciprocate towards the institution as a whole and not towards its member in particular more likely explains this wage-effort relationship.
This paper proceeds as follows: section II offers a literature review related to the current experiment. In section III, the experimental design and procedures are introduced. Section IV presents the theoretical analysis as well as behavioural prediction to formulate the hypotheses. The results of the experiment are presented in section V. Section VI provides the discussions and section VII is conclusions. The appendix contains descriptive statistics and the instructions of the experiment.
II. Literature Review
A. The concept of reciprocity
A long-standing assumption in economics portrays human beings as narrowly self-interested agents who attempt to maximize their own utility and do not care about the well-being of other human beings. However, an abundant evidence of experimental research shows that people deviate from pure selfishness in a reciprocal manner. According to Sobel (2005, p. 392): “Reciprocity refers to a tendency to respond to perceived kindness with kindness and
perceived meanness with meanness and to expect this behaviour from others.” Unlike
cooperative or retaliatory in repeated interactions where the actors anticipate future benefits from their action, reciprocity responds to a favourable or hostile action even if there are no material benefits that can be expected in the future (Fehr and Gachter, 2000). Thus, an individual with reciprocal behaviour is willing to sacrifice his/her own material payoff to either increase other’s payoff in response to kind behaviour (positive reciprocity) or decrease
6
other’s payoff in response to unkind behaviour (negative reciprocity).
However, the theory about the main driving force of reciprocal behaviour is somewhat diverse. Some researchers argue that distributional concerns and desire to maintain equity play a major role (Fehr and Schmidt, 1999; Bolton and Ockenfels, 2000). Some others believe that desire to reward kind intentions and punish unkind ones, irrespective of the payoffs distribution, is important as well (Rabin, 1993; Falk et al., 2003; Dufwenberg and Kirchsteiger, 2004; Charness and Levine, 2007). Dufwenberg and Kirchsteiger (2004) suggest that an individual should be unwilling to sacrifice any material payoff to benefit a non-volitional person. In their study of the ultimatum game, Falk et al. (2003) show that a given offer x is rejected at a higher rate if the proposer’s action signals an unfair intention compared to a situation where the proposer’s actions are determined randomly (non-intentional choices). However, some other studies show that most people still exhibit (weak) reciprocal attitudes even in situations where intention plays no role (Falk et al., 2008; Blount, 1995; and Charness, 2004).
In practice, it is not very clear which variable is more salient between intentions and distributional concerns since it is not trivial to disentangle those two. Another theory that might be insightful to explain reciprocal attitudes is that both distributional concerns and intentions matter when assessing the perceived kindness. In their theory of reciprocity, Falk and Fischbacher (2006) claim that when individuals are motivated by reciprocity considerations, they evaluate not only the outcome consequences of other’s action but also its underlying intentions. It can be seen from the results of their experiment that the same consequences of an action are perceived and reciprocated in different ways, depending on its intention. It is, therefore, as in Falk and Fischbacher (2006), the concept of reciprocity in this experiment that incorporates both the concern for the outcome as well as the underlying intention.
B. The special features of the public sector and the potential for reciprocity
The term ‘public sector’ refers broadly to the entities that are owned by the government and exist to provide various services for its citizens. To observe the reciprocity in the public
7
sector, it is useful to explore the special features of the public sector and see how they influence workers’ behaviour. However, the public sector is large, complex, and diverse across countries (Dixit, 2002). Thus, the features mentioned below might be applied partially to one institution and not all to some. However, each feature is relevant to many public sector organizations.
There are several features that make the jobs in the public sector different from those in the private sector. Unlike the jobs in the private organizations where the competition is relatively tighter, the public sector is believed to be among the most secure work environments and more stable (Postel-Vinay, 2015), especially when pay determination is based on longevity, not performance (Osborne and Gaebler, 1992). In the public sector, it is very unlikely to get payroll deductions or dismissal even when the workers are underperformed because the process of (dismissal) is so difficult and time-consuming. The system also makes it difficult to reward a good worker with a higher pay. Hence, the lack of incentives and the difficulty of getting punishment in public organizations make workers averse to exert effort (Delfgaauw and Dur, 2006) and may behave according to selfish preferences.
Regarding the wage scheme, the basic salary is publicly and centrally regulated by the government. Publicly known wage implies that every worker has a possibility to observe his/her own salary and the salary of others according to the grade they belong. Availability of pay comparison information may damage effort provision (Nosenzo, 2013) when workers perceive that they have not been paid a fair wage. The presence of salary differences between workers who perform comparable work may lead workers to believe that the wages have been set unfairly. This may ultimately reduce the worker’s willingness to exert effort.
Furthermore, the public sector has a complex hierarchical structure that consists of many levels of authority at which the ownership and control are separated. In the public sector, the government is the owner of the institutions but does not directly control the management and make executive decisions. The government typically delegates the management authority to the managers. In the context of wage setting, part of the manager’s job is to determine wage or incentive to the workers. This might be the case for the public institutions that implement
8
performance-based pay systems, in the sense that the manager gives an evaluation on the workers’ performance and decides the level of incentive the workers could get according to how well they perform (Ketelaar et al., 2007). Although the manager is the one who sets the wage, the one who pays for the wages is the owner (i.e. the government). Thus, the separation of ownership and control may have several implications. Knowing that his/her reciprocity will not benefit the manager, it may prevent worker to expend high effort. Alternatively, the wage-effort relationship may remain largely unaffected because the worker wants to reciprocate towards the one who pays for the wage (Maximiano et al., 2013).
C. Related literature on the gift exchange game
The gift exchange game (and trust game) are the most widely used games to measure reciprocity. Over decades, the game has been studied successively in many experimental researches. In its original version first developed by Fehr et al. (1993), the game was conducted using labour market framing, and it consisted of two stages. At the first stage, employers make wage proposals to workers, and then workers can either accept or reject the offer. If a worker accepts wage offered, a binding contract is concluded. If a worker rejects the offer, both players get zero. At the second stage, workers choose how much effort level to supply in response to the wage they received. Efforts are costly to workers but can be profitable for employers. The common finding in this experiment is that a worker is willing to sacrifice his/her own material payoff in exchange for the “gift” of higher wages he/she receives from the employer.
Many experimental researches extend these finding and try to test the robustness of the gift exchange game in different treatments. In the study of the gift exchange game in a multi-worker environment, Maximiano et al. (2007) have four multi-workers for each firm and find that effort levels are slightly lower than in a single worker. Relative to the pay secrecy setting, Nosenzo (2013) finds that pay disclosure can be decremental for effort provision if workers are treated unfairly. He shows that compared to pay secrecy, efforts in public wage setting are significantly lower when workers observe disadvantageous pay inequalities and can be even lower when they are in an advantageous pay inequalities, but not higher when they are
9
paid the same wages. However, some studies also find that pay comparison has weak or no effect on effort behaviour. Charness and Kuhn (2007) find that effort levels are sensitive to own wage and not affected by other worker’s wages. No pay comparison effects are also found in the real effort experiment by Hennig-Schmidt et al. (2010). Using different subject pools, Hannan et al. (2002) find that undergraduate students provide substantially less effort than MBA students.
This experiment examines worker’s reciprocal behaviour in the context of the public sector. The feature is concerned with the separation of ownership and control between the government and the manager. It also accommodates public wage schemes where workers can compare their own wage with their co-worker’s wage. This experiment is closely related to the study from Maximiano et al. (2013) which examines the robustness of the gift exchange game in a separation of ownership and control. The study concerns the fact that most firms have multi-level hierarchies in which ownership and control are separated. In the case of the gift exchange game, the owners of the firm (shareholders) do not determine the wages anymore as they delegate the authority to the manager. However, the shareholders will bear the full cost and receive the (full) profit from worker’s reciprocal behaviour. The results suggest that the gift exchange is also robust even when there is a separation of ownership and control.
Although it refers to the design by Maximiano et al. (2013), this experiment differs in some ways. While Maximiano et al. (2013) limit their study for firms, this research focuses on worker’s behaviour in the public sector. The experiment also uses different subject pools. Instead of students, real workers are used as participants. Presumably, using real workers could be a better representation in measuring worker’s reciprocity. Moreover, this study also includes the availability of pay disclosure between workers. Another distinction is that this experiment also uses different effort technology. Unlike in Maximiano et al (2013) and many other researches which commonly used an effort table, this experiment adapts the effort technology used by Nosenzo (2013) which allows for investigating both positive and negative reciprocity. Pertaining to the public wage scheme, this effort technology also makes it less complicated to compare behaviour between each worker.
10
III. Experimental Design and Procedures
A. Basic setup of the experimental game
The experiment was designed by combining the setup of the gift-exchange game from Nosenzo (2013) and Maximiano et al. (2013). The basic setup of the experiment concerns an employer which employs two workers. At the first stage, the employer is endowed 22 points from which he/she pays a wage to the two workers. The wages can take three values: 1 point, 4 points, or 7 points. At the second stage, the workers then indicate the effort level after observing their own wage and the wage of their co-worker. The workers can choose among three levels of effort: low, medium, or high. Low effort costs the worker 1 point and reduces the employer’s earnings by 4 points. Medium effort costs the worker nothing and leaves the employer’s earnings unchanged. High effort costs the worker 1 point and increases the employer’s earnings by 4 points. Unlike the effort technology commonly used in the gift exchange game where workers can only choose whether or not to reward the employer, this effort technology allows for both positive reciprocity (rewarding when workers exert high effort) and negative reciprocity (punishing when workers exert low effort).2 The combination of wage and effort determine the monetary payoffs for each player, which were given by:
Employer : 22 + 4 (e1 + e2) – w1 – w2
Worker : wi – ei
where e denotes the worker’s effort, and w represents the wages offered.
The game consists of three treatments. The Baseline treatment follows the standard trilateral gift exchange game in which there is a single manager as the owner who determines wages for two workers. Both workers then choose an effort level after observing the co-worker’s wage. Treatment 1 considers the hierarchical structures that separate ownership and control, meaning that the one who owns the institution is different with the one who controls it. In this treatment, the role of government as the owner of the institution is introduced, resulting
11
the gift exchange game with four players involved. However, as the owner, the government is not responsible for determining the wage anymore. This responsibility has been delegated to the manager as part of his/her job in controlling the institution. The government, therefore, does not make any decision at all. Separation of ownership and control implies that the manager is not a full residual claimant (Maximiano et al., 2013) i.e. the manager does not bear the full wage costs and thus does not receive the full benefit from his/her decision. Instead, the government claims the entire benefits as it is the one who bears the full wage costs.
Treatment 2 takes into account the possibility that workers may not care about control at all and only care about the fact that they are being paid. It is, therefore, unlike Treatment 1 where the wage is determined endogenously by a member of the institution, Treatment 2 considers the wage decision by an exogenous party. Instead of being determined by the manager, the wage is randomly determined by the experimenter, with each possible wage combination being equally likely. This treatment also portrays the natural setting of the public sector environment in which the wage composition was set exogenously by a regulation outside the member of institutions. Table 1 provides the overview of the three treatments used in the experiment.
Table 1. Overview of the Treatments and Monetary Payoffs
Treatment Player Wage
deter-mined by Government’s earnings Manager’s earning Worker’s earnings Baseline 1 employer 2 workers Manager - 22 + 4 (e1 + e2) – w1 – w2 wi – ei Treatment 1 1 government 1 manager 2 workers Manager 22 + 4 (e1 + e2) – w1 – w2 0 wi – ei Treatment 2 1 government 2 workers Exogenously 22 + 4 (e1 + e2) – w1 – w2 - wi – ei
12
In this experiment, the participants played the game only once, rather than repeatedly in a row. The advantage of using one-shot interaction is that subjects cannot build any reputations or learning effects. Moreover, repeating the game does not guarantee the collection of sufficient data from observations for all possible information needed in the game. To overcome the shortcomings, the strategy method is used, where each subject has to formulate a complete strategy for all possible situations. Each worker, therefore, has to make nine contingent effort choices for all nine possible wage combinations offered. This has advantages to elicit wage-effort relationship for each individual and all possible wage combinations, thus enabling the generation of a relatively large amount of data at low cost. Although there is a possibility that using the strategy method may generate different results than using the direct response method, most experimental studies show that the strategy method should not be a matter, e.g. Brandts and Charness (2011) who directly compare the direct response method and strategy method find that the two elicitation methods do not prompt to qualitatively different results. Moreover, in the context of the gift exchange game experiment, Gachter and Thoni (2010) show that both the direct response method and the strategy method yield consistent results.
B. The experimental procedures
The participants in this experiment were the employees of the Ministry of Finance of the Republic of Indonesia. Subjects were recruited by distributing an anonymous link to potential participants through various social media, such as Facebook Messenger, WhatsApp, and email. The average age of participants was 28.6 years and 65.6% of them were male. The year of service ranged from 2 years until 15 years with the average of 7.2 years.3 Overall, the experiment was conducted in four sessions with a total of 112 subjects. One session was run to gather employer’s decision (16 subjects), while three sessions were assigned for workers (each session consisted of 32 subjects). The game in Session 1 consisted of three parts which represented three treatments. All subjects who played the role of an employer (either manager or government) played for three consecutive parts (all treatments), and then matched with
13
two workers who made decisions in Session 2 - 4 a few days later. On the other hand, subjects who played the role of workers in each session only played once so that no subject took part in more than one session. In particular, an employer was matched with two different workers in each treatment. Thus, it ensured that subjects were never matched with the same person. This randomization was used for the sake of efficiency due to limited resources. Since the main interest of this experiment is to study worker’s behaviour, this randomization should not be an issue and could be useful as a way to get more data from observations.
The experiment was conducted online using Qualtrics software. Subjects started the session by reading general and detailed instructions of the experiment.4 Apart from these instructions, subjects who played the role of an employer were given an additional instruction. To ensure that subjects understood the experiment, all subjects had to solve a set of control questions correctly before the experiment started. In particular, they were asked to calculate the payoff for each different role to make sure all subjects understood how the manager’s wage decision affects the government’s payoff and how the worker’s effort decision affects the manager and the government differently. The experiment was held in a separate time for the employer and the workers. At the first stage, an online survey was spread to collect decisions from employers for all treatments. In the Baseline session, all subjects performed the role of owner-manager and had to set the wages for both workers. In Treatment 1, half of the subjects were given the role of government, and the remaining half were assigned the role of manager. Note that the government took no decision at all. Therefore, for the sake of efficiency, the role was switched so that eventually each subject made decisions. Meanwhile, in Treatment 2, all subjects played the role of government. Since the wages were randomly determined by the experimenter, the subjects did not make any decision.
In the second stage, a few days later, subjects who played the role of workers were matched with employers from the previous stage. For each session, half of the subjects were assigned as Worker A and the remaining half were assigned as Worker B. When making actual choices, workers had to indicate their effort choices for all nine possible wage scenarios
14
without knowing the actual wage set by the manager (or set by the experimenter in Treatment 2). At the end of the experiment, subjects completed a short background questionnaire. The number of points a subject had earned were converted into rupiahs, with 1 point equivalent to Rp10,000. Three groups of players corresponding to one employer and two workers in each treatment (9 participants in total) were randomly selected to receive payment. Subjects were not informed of which situation was actually relevant and did not learn their own and other member’s earnings unless their decisions were chosen for payment.
IV. Theoretical Analysis and Behavioural Predictions
Under the standard economics assumptions of self-regarding preferences and according to the subgame perfect Nash equilibrium, a purely self-interested and payoff maximizing worker chooses minimum effort level, irrespective of the actual wage offered. In that case, a medium effort will be chosen for every wage offered in all three treatments. A rational and selfish employer, therefore, will offer the lowest wage in order to minimize the costs. Note that according to standard economics predictions, the availability of pay comparison information will not influence the effort choices. Thus, workers choose effort that minimizes costs, regardless of the wage paid to the co-workers.
Contrary to the standard predictions, a large number of experimental study find a significant positive wage-effort relationship, indicating a strong evidence of apparent reciprocity between workers and employer (Fehr et al., 1993; Fehr et al., 1998; Brandts and Charness, 2004; Charness, 2004; and Falk, 2007). Workers are likely to exert high effort in response to a high wage offered by the employer, even in an anonymous one-shot interaction where there are no benefits that can be expected in the future. These findings support the efficiency wage theory by Akerlof (1982) which posits that firms will offer a higher than market clearing wage, expecting a higher effort in return. Therefore, a positive own wage-effort relationship is also predicted to be robust in this experiment.
Furthermore, workers evaluate employer’s actions with a norm they consider fair and decide whether or not to supply a higher effort. However, workers’ fairness consideration may
15
introduce a potential drawback for the employer because what constitutes as a fair wage may not only depend on one’s own wage but also on relative pay comparison (Charness and Kunt, 2007; Nosenzo, 2013). When choosing an effort, workers not only look into how much wage they receive but also compare their own wage with the wage of their co-worker (if the information is available). If workers believe that wages have been set unfairly i.e. there is a salary gap among workers who perform comparable work, it will reduce job satisfaction and eventually prevent them to expend higher effort (Nosenzo, 2013). Furthermore, in their theory of reciprocity, Falk and Fischbacher (2006) suggest that in addition to distributional concerns the employer’s intentions also matter. Thus, when workers have a preference to behave reciprocally, they take into account the kindness of an employer’s action and evaluate not only the outcome consequences but also the underlying intentions.
In the Baseline treatment, the owner-manager decides the wage by him/herself and obtains the entire profits. A manager who offers a high wage will benefit the workers and can be considered kind, especially when the manager has to bear the entire wage cost him/herself. On the contrary, a low wage can be considered unkind. Therefore, it is expected that when being underpaid i.e. when the worker is paid a lower wage than his/her co-worker, a reciprocal worker will choose low effort (less rewarding and more punishment), and when being overpaid i.e. when the worker is paid a higher wage than co-worker, he/she will choose high effort (more rewarding and less punishment). This hypothesis corresponds to the motivation to reward (punish) the one who controls the wage, motivation to reward (punish) the one who pays for the wage, or a combination of both because the owner-manager in this treatment represents both the one who controls and pays the wage. Therefore, the predictions still hold no matter which of these roles the worker prefers to reward (or punish).
In Treatment 1, however, the ownership and control are separated. The manager who determines the wage is not the owner who pays the wage costs and thus does not receive benefits. Although a higher wage offered by the manager still can be considered kind, the worker has no opportunity to reward (punish) the manager’s kind (unkind) intention for setting the wage. Exerting high (low) effort has no effect on the manager’s earnings but increases (decreases) the government’s earnings. In that case, it is expected that the worker
16
chooses the medium effort level, regardless of wages offered. However, the results from Maximiano et al. (2013) suggest that the workers do not care who and how the wage is determined. They only care about those who pay for the wage. In that case, if the workers feel reciprocate towards the government as the one who pays for their wage, the prediction will be the same as the Baseline. This prediction also holds when the worker perceives the whole institution as his/her employer, regardless of its composition. To put it differently, the worker feels reciprocate towards the institution as a whole and not towards one of its members in particular. The worker, therefore, is expected to behave like in the Baseline treatment as well.
In Treatment 2, the wage is randomly determined by the experimenter and thus it eliminates the controlling role from internal parties. When an employer does not have the authority to choose the wage (a condition where there is no intention), the worker should be unwilling to sacrifice any material payoff to benefit a non-volitional employer (Dufwenberg and Kirchsteiger, 2004). Therefore, if the worker is mainly motivated by reciprocity towards the one who pays for the wage, the prediction will be the same as in the Baseline and Treatment 1. Another possibility that can explain the worker’s behaviour in this treatment is that he/she may consider both control and ownership as an important determinant for his/her reciprocal feeling, although he/she does not care how these two variables are parcelled out within the institution (Maximiano et al., 2013). In that case, the worker also chooses low effort when underpaid (less rewarding and more punishment) and chooses high effort when overpaid (more rewarding and less punishment), but the proportion of reciprocal workers will be smaller compared to the Baseline.
V. Results
This section will highlight the results of the experiment. Since the aim of this experiment is to examine whether workers exhibit any reciprocal attitudes in the public sector environment, the results, therefore, will begin by comparing the worker’s behaviour across all treatments. In particular, it compares the effort chosen by workers in the Baseline with effort chosen in Treatment 1 and Treatment 2.
17 1. Baseline vs. Treatment 1
The effort technology used in this experiment allows both rewarding and punishing the employer through positive and negative reciprocity. The results, therefore, will focus on low effort and high effort chosen by workers. Figure 1 presents the proportions of low effort and high effort choices made by workers in the Baseline and Treatment 1.
* The Figure excludes the category of medium effort, thus % high effort + % low effort = 100% - %
medium effort. Bars are based on effort choices made by 32 workers in the Baseline and Treatment 1.
In general, Figure 1 suggests that in both treatments the higher the wage paid to the workers, the higher the effort they supplied. Only a small fraction of workers rewards the employer with high effort when they are paid 1 point, while the proportion of high effort choices when they are paid 7 points varies from 50% to 84% depending on treatment and relative pay comparison. On the contrary, workers rarely punish the employer with low effort when their own wage is 7 points, but a 1-point wage triggers punishment from 10% up until 50% of the time. These results contradict the prediction of standard economics theory that workers will always exert medium effort level irrespective of the wage received. Thus, as in many other gift exchange experiments, the positive own wage-effort relationship also holds in this
60% 40% 20% 0% 20% 40% 60% 80% 100% 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts
1 point 4 points 7 points 1 point 4 points 7 points
Figure 1. Low and high effort rates:
Baseline vs. Treatment 1*
high effort low effort
BASELINE TREATMENT 1
Co-worker’s wage Own wage
18 experiment.
Comparing the results from the Baseline and Treatment 1, Figure 1 shows that workers’ behaviour appears very similar across treatments. In both treatments, underpaid workers (i.e. a worker who is paid 1 point while his/her co-worker’s wage is 4 or 7 points, and a worker who is paid 4 points while his/her co-worker’s wage is 7 points) infrequently reward the employer with high effort, whereas punishment rates i.e. the proportion of workers choosing low effort vary from 22% to 47% in the Baseline and 25% to 50% in Treatment 1. Similarly, when workers are overpaid (i.e. when worker’s own wage is 4 points while his/her co-worker’s wage is 1 point, and when co-worker’s own wage is 7 points while his/her co-co-worker’s wage is 1 point or 4 points), there are virtually no low effort chosen in both treatments. Instead, reward rates are between 62% and 81% in the Baseline, not much different from Treatment 1 which ranged 56% to 84%. A series of chi2-tests for equality of proportion also shows that differences in reward rates or punishment rates between the Baseline and Treatment 1 are insignificant for all combination of pay differentials (all p > 0.106). Finally, there seems to be small differences in punishment rates between two treatments when workers are paid the same wage as the co-worker, i.e. when both workers are paid 1 point. The difference in punishment rates between the Baseline (9.4%) and Treatment 1 (31.3%) are found to be statistically significant (p = 0.029).5 Yet, the result fails to find significant differences between treatments for punishment rates when both workers are paid a 4-point wage or 7-point wage (p > 0.162), as well as for reward rates in any combination under the same wage (all p > 0.127).
To further explore these two observations, a regression analysis is conducted using dummies for the treatments, as well as dummies for the own wage and dummies for the co-worker’s wage. The regression also estimates the random effect generated from the panel data (9 data
5Considering that chi2-test only finds a significant difference between the Baseline and Treatment 1 when
workers are paid the same 1-point wage, a power analysis is conducted to account for the effect size. Since the likelihood of obtaining statistically significant results increases as the sample size increases, there is a possibility that significant result could be detected with more observations. With the power analysis, the results find an estimate sample sizes required for each treatment is 41 subjects (alpha 5%, power 80%).
19
points per individual worker), as well as controls the background characteristics of the sample such as gender, age, and year of service. The results are presented in Table 2. Generally, the robustness check shows that the differences in effort level between the Baseline and Treatment 1 are insignificant for all wage combinations.
Table 2. Random effect linear regression on effort level in Treatment 1
Coefficient Robust standard error
Treatment 1 .0039 .0538 Own wage .1398 .0106 Co-worker’s wage -.0608 .0112 Gender .0724 .0562 Age .0286 .0204 Year of service -.0091 .0191 Constant 1.1631 .5264
Overall, the results in both treatments find a positive own wage-effort relationship. As for relative pay comparison, the results in the Baseline treatment support the hypothesis that most of the workers expend high effort when being overpaid and choose low effort when being underpaid. Considering at whom the workers like to reciprocate, the results for Treatment 1 clearly reject the hypothesis that workers will choose medium effort level regardless of the wage offered. This indicates that the workers do not only reciprocate towards the one who controls the wage. Certainly, the one who pays for the wage plays a role to explain worker’s reciprocal behaviour. This holds because the results show no significant differences between the wage-effort relationship in the Baseline and Treatment 1. These findings also suggest that workers may reciprocate towards the institution as a whole and do not really care about its composition. Apparently, workers only care about the fact that they are being paid; whether by the (owner) manager or the government is not important.
20 2. Baseline vs. Treatment 2
This section will compare the results for Treatment 2 with the Baseline. Figure 2 displays the proportion of low effort and high effort choices made by all workers. As in the Baseline and Treatment 1, the results in Treatment 2 also suggest that higher own wage levels increase the proportion of high effort chosen and decrease the frequency of low effort. This result also indicates that there is a positive own wage-effort relationship.
As seen in Figure 2, there are slight differences between the results from the Baseline and Treatment 2. In particular, when workers are overpaid relative to their co-worker, the reward rates are slightly higher in the Baseline than in Treatment 2, i.e. when the own wage is 7 points and co-worker’s wage is 4 points, reward rates in the Baseline are 75% while reward rates in Treatment 2 are a bit lower at 59%. Also, when the own wage is 4 points and co-worker’s wage is 1 point, reward rates in Treatment 2 are 12.5% smaller than the Baseline.6
It is very likely that non-intentional action reduces worker’s reciprocal behaviour in Treatment 2. Moreover, the proportion of workers choosing low effort when being overpaid is the same for any wage comparisons both in the Baseline and Treatment 2. In contrast, when being underpaid (e.g. when the own wage is 1 point or 4 points and co-worker’s wage is 7 points), workers in the Baseline punish at the higher rate than in Treatment 2.7 This might be the case because in Treatment 2 wages are randomly chosen by the external party and thereby it decrements the worker’s willingness to punish the non-volitional person (in fact, workers do not have an opportunity to punish (reward) the one who set the wage). Reward rates do not seem to vary markedly between the Baseline and Treatment 2 when workers are underpaid. However, there seems to be no other significant differences found in punishment rates between treatments for all combinations of the same wages.
6Using chi2-test for equality of proportion, the results do not find significant differences between treatments (p
> 0.183) though.
21
* The Figure excludes the category of medium effort, thus % high effort + % low effort = 100% - % medium effort. Bars are based on effort choices made by 32 workers in the Baseline and Treatment 2.
Table 3 reports the regression analysis of effort behaviour in the Baseline and Treatment 1. Similarly to the analysis in the previous sub-section, the regression estimates using dummies for the treatments, as well as dummies for the own wage and dummies for the co-worker’s wage. The result shows no significant differences in effort level between the Baseline and Treatment 2. With respect to analysis from Figure 2, the negative coefficient also supports the prediction that the treatment effect lowers the proportion of workers being reciprocal in Treatment 2 compared to the Baseline as the workers should be unwilling to reward (punish) a non-volitional employer.
Overall, the results in Treatment 2 do not differ significantly with the results in the Baseline. Since the controlling role is entirely absent in Treatment 2, the results are in line with the prediction that workers are only concerned about the one who pays for the wage. Although it is not statistically significant, the reward rates (when workers are overpaid) and punishment rates (when workers are underpaid) in Treatment 2 are in fact smaller than in the Baseline. This indicates that control per se is important for the reciprocal relationship (Maximiano et al., 2013; Charness, 2004). 60% 40% 20% 0% 20% 40% 60% 80% 100% 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts 1 p o in t 4 p o in ts 7 p o in ts
1 point 4 points 7 points 1 point 4 points 7 points
Figure 2. Low and high effort rates:
Baseline vs. Treatment 2*
high effort low effort
BASELINE TREATMENT 2
Co-worker’s wage Own wage
22
Table 3. Random effect linear regression on effort level in Treatment 2
Coefficient Robust standard error
Treatment 2 -.0666 .0593 Own wage .1241 .0103 Co-worker’s wage -.0521 .0109 Gender .0020 .0582 Age -.0070 .0204 Year of service .0414 .0236 Constant 2.013 .4717 3. Employer’s behaviour
Although it is not the main interest of the experiment, the section will briefly discuss the overview of (owner-) managers’ behaviour. According to standard economic assumptions, the owner-manager should pay minimum wage in the Baseline in order to maximize own payoffs. However, if the owner-manager expects that workers will exhibit any positive or negative reciprocal behaviour in response to the wage offered, he/she may pay more than minimum wage. The manager may consider relative pay comparison as well and choose to pay the fair wage for both workers. However, in Treatment 1 where the manager who sets the wage does not benefit from the worker’s effort, any wage offered from the manager is consistent with selfish preferences. If the manager offers the same high wage for both workers, it may indicate that he/she has a preference for equality. Since wages in Treatment 2 are determined randomly, the analysis will focus on the Baseline and Treatment 1.
Table 4 shows the wage distribution among workers for six possible wage combinations.8
8Since any possible combination of unequal wage between two workers does not influence the
manager/government’s earnings, the data presented does not distinguish the wage combination between Worker A and Worker B. For example, combination of 1-point wage to Worker A and 4-point wage to Worker B will be treated the same as combination of 4-point wage to Worker A and 1-point wage to Worker B.
23
Although it should be the best response, there is no manager who offers the combination of the minimum wage (i.e. offers 1 point wage for both workers) in the Baseline and Treatment 1. Typically, the manager sets a higher wage of either 4 points or 7 points in most cases. This is consistent with the efficiency wage hypothesis (Akerlof, 1982) that an employer is willing to pay a higher wage with the expectation of a higher effort provided by the workers.
Table 4. Wage distribution in the Baseline and Treatment 1
Wage combination Percentage offers Baseline Treatment 1 1 point 1 point 0,00% 0,00% 4 points 4 points 31,25% 31,25% 7 points 7 points 18,75% 12,50% 1 point 4 points 0,00% 12,50% 1 point 7 points 6,25% 0,00% 4 points 7 points 43,75% 43,75%
In both treatments, half of the managers choose to pay a fair wage for both workers: 31.25% of the managers pay an equal wage of 4 points in both treatments, and the percentage of managers who pay an equal wage of 7 points is 18.75% in the Baseline and 12.5% in Treatment 1. This implies that the manager has equality preferences even though in Treatment 1 he/she does not receive the benefits. However, the majority of the managers (43.75%) prefer to pay a combination of 4 points and 7 points in both treatments. This indicates that the manager does not care about equality. The managers probably think that the workers may tend to provide a medium effort if the manager gives both workers the maximum wage since they learn about their co-worker's wage. If the managers provide one of the workers a high wage and a medium wage to another worker, there is a possibility that the worker who got a high wage will provide high effort.
24
VI. Discussion
As in many previous gift exchange experiments, this research also finds a positive wage-effort relationship in all treatments. Typically, overpaid workers exhibit positive reciprocal behaviour by rewarding high wage with high effort while underpaid workers show negative reciprocity by punishing low wage with low effort. Most importantly, these results also hold in Treatment 1 where ownership and control is separated. Apparently, most of the workers still act reciprocally even though the manager who controls the wages does not receive benefits from their reciprocal behaviour. This indicates that workers simply care about the fact that the government pays them. The behaviour also can be derived by a motivation to reciprocate the institution as a whole, i.e. workers are not really concerned about how ownership and control are divided within the institution.
Similar to the Baseline and Treatment 1, the results also exhibit positive and negative reciprocity among workers in Treatment 2. Indeed, workers behave reciprocally towards the one who pays for the wage even when the controlling role is absent. The fact that exogenously given wages lower the proportion of reciprocal workers in Treatment 2 relative to the Baseline indicates that workers perceive both ownership and control as important determinants in the reciprocal relationship and do not differentiate these two roles in particular. It also implies that workers are reciprocal towards the institution as a whole.
These finding support the results from Maximiano et al. (2013) that worker’s motivation to reciprocate does not diminish even when the one who sets the wages does not receive the benefits. In this experiment, however, the behaviour observed may appear stronger than in Maximiano et al. (2013) since it accommodates worker’s negative reciprocity toward employer. When being treated unfairly, the worker has an opportunity to punish the employer by providing low effort which eventually decreases employer’s profits. Moreover, this paper adds the availability of pay comparison to further explore whether the separation of ownership and control also robust when worker can observe not only his/her own wage but also the co-worker’s wage. Somewhat surprisingly, the wage-effort relationship does not differ between treatments even in the presence of pay disclosure. This implies that both in
25
the case of absolute wage and relative wage, the result supports the hypothesis that workers only care about the one who pays for the wage (owners) and the hypothesis that workers reciprocate toward the institution as a whole, not toward its member in particular.
Besides, there were some interesting yet unexpected decisions made when the workers were paid the same 1-point wage as their co-worker. In all treatments, notably in the Baseline and Treatment 1, several workers had chosen a high wage even though they were paid a low wage. A possible explanation is that these workers probably tried to impress the manager by giving high effort even when their wage is low with an expectation that the manager will increase their wage in the future. Even in the one-shot interaction where no future benefits can be expected, the reputation formation can happen especially when subjects are real workers who are sensitive to some similar situations in work life. Conversely, when being paid the same 7-point wage as their co-worker, there were individuals who had chosen low effort. As the proportion is relatively low, this decision may be made by mistakes or subjects may not completely understand the experiment.
Another unexpected finding is that there are five subjects (i.e. one subject in the Baseline and four subjects in Treatment 1) who always chose high effort regardless of the wage offer and relative pay comparison. These subjects may behave beyond reciprocity and could be altruistic because they are unselfish and are only concerned about the well-being of others. On the other hand, eight subjects (i.e. one subject in the Baseline, three subjects in Treatment 1 and four subjects in Treatment 2) always chose medium effort regardless of the wage offer. This indicates that the subjects behave according to self-regarding preferences. They are unwilling to sacrifice any material payoffs for others and only care to maximize own payoffs.
Interpreting these results, some limitations should be considered. First, the experiment was conducted using an online survey. Compared to the laboratory setting, online experiments may have weaker experimental controls that ensure reliability and internal validity (Anderhub, 2001). For example, subjects may ignore or read the instructions too carelessly, leading to lower quality of data. Second, this experiment is highly contextualized. The game is framed in a labour market setting which closely resembles the public sector environment.
26
Since the subjects are public servants, it may make them play according to some situations they might be familiar with. For instance, some workers in the experiment may always choose a high wage out of habit because they expect managers to benefit somewhere in the future and ultimately increase their wage, as it commonly happens in real life. The third limitation, and somewhat related point, concerns the one-shot interaction of the game. As one-shot interaction limits the scope for reputation and learning, it remains an open question whether workers would still reciprocate a high wage with a high effort if the experiment is conducted in repeated interactions. It would be more insightful for future research to conduct an experiment in which subjects play the games a number of times in a row and see whether workers’ behaviour changes.
The result contributes to the experimental literatures on the gift exchange game within multi-level hierarchies in which ownership and control are separated. This finding also contributes to the implementation of wages policy in the public sector. Although workers apparently reciprocate toward the institutions as a whole, workers are more incline to reciprocate when the endogenous party sets the wages rather than when wages are determined exogenously. This implies that the implementation of performance-based pay in which the manager involves in determining wages or incentives might be a better way to motivate workers to expend more effort, compared to when wages are set by regulation.
VII. Conclusions
This paper presents an experimental investigation on how the features of public sector influence workers’ reciprocal behaviour. In particular, it investigates workers’ reciprocity when there is a separation of ownership and control between the government and manager, and, at the same time, the wages are publicly known. Although some of the results could not be statistically confirmed, all findings indicate that the gift exchange game in this experiment also appears to be robust to the feature of public sector environment in which there is a separation of ownership and control of public wage.
27
even when there is a separation of ownership and control within the public sector institution. Workers are likely to reward a high wage with high effort and punish a low wage with low effort even when the one who sets the wages does not share the benefit from their reciprocal behaviour. It also suggests that workers feel reciprocate towards the institution as a whole, regardless of its composition.
28
VIII. References
Akerlof, G. A. (1982). Labor contracts as partial gift exchange. The Quarterly Journal of Economics, 97(4), 543-569.
Benndorf, V., & Rau, H. A. (2015). Fairness and Learning in Multi-Employee Gift-Exchange Games: An Experimental Analysis.
Blount, S. (1995). When social outcomes aren′t fair: The effect of causal attributions on preferences. Organizational behavior and human decision processes, 63(2), 131-144. Bolton, G. E., & Ockenfels, A. (2000). ERC: A theory of equity, reciprocity, and competition. American economic review, 166-193.
Brandts, J., & Charness, G. (2004). Do labour market conditions affect gift exchange? Some experimental evidence. The Economic Journal, 114(497), 684-708.
Brandts, J., & Charness, G. (2011). The strategy versus the direct-response method: a first survey of experimental comparisons. Experimental Economics, 14(3), 375-398.
Charness, G. (2004). Attribution and reciprocity in an experimental labor market. Journal of labor Economics, 22(3), 665-688.
Charness, G., & Kuhn, P. (2007). Does pay inequality affect worker effort? Experimental evidence. Journal of labor economics, 25(4), 693-723.
Charness, G., & Levine, D. I. (2007). Intention and stochastic outcomes: An experimental study. The Economic Journal, 117(522), 1051-1072.
Delfgaauw, J., & Dur, R. (2008). Incentives and workers’ motivation in the public sector.
The Economic Journal, 118(525), 171-191.
Dixit, A. (2002). Incentives and organizations in the public sector: An interpretative review.
Journal of human resources, 696-727.
Dufwenberg, M., & Kirchsteiger, G. (2004). A theory of sequential reciprocity. Games and economic behavior, 47(2), 268-298.
Falk, A., Fehr, E., & Fischbacher, U. (2003). On the nature of fair behavior. Economic Inquiry, 41(1), 20-26.
Falk, A., & Fischbacher, U. (2006). A theory of reciprocity. Games and economic behavior, 54(2), 293-315.
29
Falk, A., Fehr, E., & Fischbacher, U. (2008). Testing theories of fairness—Intentions matter. Games and Economic Behavior, 62(1), 287-303.
Fehr, E., Kirchsteiger, G., & Riedl, A. (1993). Does fairness prevent market clearing? An experimental investigation. The quarterly journal of economics, 108(2), 437-459.
Fehr, E., Kirchsteiger, G., & Riedl, A. (1998). Gift exchange and reciprocity in competitive experimental markets. European Economic Review, 42(1), 1-34.
Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. The quarterly journal of economics, 114(3), 817-868.
Fehr, E., & Gächter, S. (2000). Fairness and retaliation: The economics of reciprocity. The journal of economic perspectives, 14(3), 159-181.
Gächter, S., & Thöni, C. (2010). Social comparison and performance: Experimental evidence on the fair wage–effort hypothesis. Journal of Economic Behavior & Organization, 76(3), 531-543.
Hannan, R. L., Kagel, J. H., & Moser, D. V. (2002). Partial gift exchange in an experimental labor market: Impact of subject population differences, productivity differences, and effort requests on behavior. Journal of Labor Economics, 20(4), 923-951.
Hennig‐Schmidt, H., Sadrieh, A., & Rockenbach, B. (2010). In search of workers' real effort reciprocity—a field and a laboratory experiment. Journal of the European Economic
Association, 8(4), 817-837.
Ketelaar, A., Manning, N., & Turkisch, E. (2007). Performance-based arrangements for senior civil servants OECD and other country experiences. OECD Working Papers on Public
Governance, (5), 0_1.
Maximiano, S., Sloof, R., & Sonnemans, J. (2007). Gift exchange in a multi‐worker firm. The Economic Journal, 117(522), 1025-1050.
Maximiano, S., Sloof, R., & Sonnemans, J. (2013). Gift exchange and the separation of ownership and control. Games and Economic Behavior, 77(1), 41-60.
Nosenzo, D. (2013). Pay secrecy and effort provision. Economic Inquiry, 51(3), 1779-1794. Osborne, D., & Gaebler, T. (1992). Reinventing government: How the entrepreneurial spirit is transforming government. Reading Mass. Adison Wesley Public Comp.
Postel-Vinay, F. (2015). Does it pay to be a public-sector employee?. IZA World of Labor. Rabin, M. (1993). Incorporating fairness into game theory and economics. The American economic review, 1281-1302.
30
Sobel, J. (2005). Interdependent preferences and reciprocity. Journal of economic literature, 43(2), 392-436.
31
Appendix A: Descriptive Statistics
The descriptive statistics about background characteristics of the subjects in this experiment are presented in Table 5 below.
Table 5. Descriptive statistics for subjects
Treatment Condition
Gender Age Year of services
% Male % Female Min Max Mean Min Max Mean
Statistic for workers (32 subjects in each treatment)
Baseline 71,87% 28,13% 22 30 27,88 2 9 6,66
Treatment 1 68,75% 31,25% 25 38 28,19 3 15 6,66 Treatment 2 65,63% 34,37% 26 38 29,90 5 14 8,43 Statistic for employers (16 subjects)
All
Treatments
32
Appendix B: Instructions for Treatment 1
In the following is general instructions and detailed instructions used for Treatment 1, as well as additional instruction for the employers.
GENERAL INSTRUCTIONS
Welcome!
You are about to take part in an experiment to study economic decision making. There are other people who are also participating in this experiment. During the experiment, you and the other participants will be asked to make decisions. It is possible to earn money in doing so, therefore it is very important to follow the instructions and make the decisions carefully. Participants:
There are three types of participants: governments, managers and workers, which will be referred as a group. You will be randomly assigned to one of these roles. Which role you have, you will be notified at the start of the experiment. You will not be informed about who of the other participants are in your group, either during or after the experiment. Therefore, all decisions are made anonymously.
Duration:
This experiment will take approximately 10 minutes. Earnings:
Your earnings will depend partly on your decisions and partly on the decisions of the other participants in the experiment. Earnings will be calculated in points. All points that you gain during the course of the experiment will be exchanged into rupiahs. The exchange rate will be:
1 point = Rp10,000.00
Only one group of participants in the experiment will be randomly selected for payment. If your decision is chosen for payment, you will be contacted by the experimenter to claim your earnings. Since each decision can be chosen for payment, it is important that you treat each decision carefully.
33
DETAILED INSTRUCTIONS
In this experiment, you are taking part in a study of the labour market. You will be randomly matched with three other participants to form a group. Each group consists of four people: one participant takes the role of Government, one participant takes the role of Manager, one participant takes the role as Worker A, and one participant takes the role of Worker B. You have to make a decision without knowing with whom you are matched. If you represent a government, you actually do not have to make any decisions at all. If you are a manager, you have to set the wage of the worker. If you are a worker, you have to decide which effort level you want to provide for each possible wage set by the manager.
The structure of the decision-making within each group is as follows.
• The Government is initially endowed with 22 points from which he/she pays a wage to the two workers with whom he/she is matched.
• Wages can take three values: 1 point, 4 points or 7 points.
• However, the Government delegates the authority of determining the wages to the Manager.
• The Manager then chooses wages for two workers.
• The wage the Manager chooses for Worker A and Worker B will be subtracted from the Government’s 22 points endowment.
• Each Worker is then informed of the wages offered by the Manager, i.e. workers learn their own wage and their co-worker’s wage.
• Each Worker chooses then independently and in private an effort level: LOW (-1), MEDIUM (0) or HIGH (+1).
o Low effort costs the worker 1 point and reduces the Government’s earnings by 4 points.
o Medium effort costs the worker nothing and leaves the Government’s earnings unchanged.
o High effort costs the worker 1 point and increases the Government’s earnings by 4 points.
• The combination of wage and effort determine the earnings for each player, as given below:
o The Government has 22 points initial endowment and incurs costs from the wages paid to the two Workers. The costs are the sum of the two wages the Government pays to the Workers. The Government then receives revenue
34
from the effort chosen by the two Workers. The revenue produced by each Workers equals 4 times the effort he/she chooses. The Government’s earnings are therefore:
Government’s
earnings =
22 – Wage offered to Worker A – Wage offered to Worker B + 4 * (Effort level of Worker A + Effort level of Worker B)
o The Manager sets the wages for both workers but does not bear cost of wages, thereby does not receive generated benefits. Thus, the Manager's earnings is 0 (zero).
o The Worker receives the wage from the Government as revenue, and may incur an effort cost. The medium effort is costless, while low or high effort costs 1 point. Therefore, the Worker's earnings are:
Worker’s earnings = Wage – Cost of effort
On the next pages you will find a hypothetical example which will illustrate how to calculate the earnings of each member in the group.
Hypothetical Example for Demonstration Purposes
Suppose the Manager chooses a 1-point wage to Worker A and a 4-point wage to Worker B. Suppose that the Worker A chooses LOW effort and the Worker B chooses HIGH effort. This situation results in the following earnings:
GOVERNMENT’S EARNINGS:
The Government pays a total of 5 points: 1 point to Worker A and 4 points to Worker B. The Government receives a 4 points revenue from the HIGH effort of Worker B, but Worker A’s LOW effort choice decreases his/her earnings by 4 points.
Therefore, the Government’s earnings are: 22 – 4 – 1 + 4 – 4 = 17 points. MANAGER’S EARNINGS:
The Manager receive nothing (0 points) regardless of the wages he/she chooses and workers’ effort level.
WORKER A’S EARNINGS:
Worker A receives a 1-point wage and chooses low effort level. LOW effort costs 1 point to the worker.
35 Therefore, Worker A’s earnings are: 1 – 1 = 0 points. WORKER B’S EARNINGS:
Worker B is paid a 4-point wage. HIGH effort costs 1 point to the worker. Therefore, Worker B’s earnings are: 4 – 1 = 3 points.
Although the structure of the decision-making within each group is the one described above, in this experiment Workers make their decisions before learning the wages that the Manager has actually chosen.
If you are a Worker, you are asked to take a decision for each possible situation that may arise. You have to indicate what you would do in each of the following NINE SITUATIONS: 1. The Manager chooses a 1-point wage for you and a 1-point wage for another Worker. 2. The Manager chooses a 1-point wage for you and a 4-point wage for another Worker. 3. The Manager chooses a 1-point wage for you and a 7-point wage for another Worker. 4. The Manager chooses a 4-point wage for you and a 1-point wage for another Worker. 5. The Manager chooses a 4-point wage for you and a 4-point wage for another Worker. 6. The Manager chooses a 4-point wage for you and a 7-point wage for another Worker. 7. The Manager chooses a 7-point wage for you and a 1-point wage for another Worker. 8. The Manager chooses a 7-point wage for you and a 4-point wage for another Worker. 9. The Manager chooses a 7-point wage for you and a 7-point wage for another Worker. Please note that one of these situations will be taken into account to determine your and the other member’s earnings, so make your choices carefully.
You will be informed of which situation is actually relevant if your group is chosen for payment. The wages the manager has actually chosen will determine, for each worker, which of the nine situations above (1 to 9) counts for the computation of earnings. The Worker’s choices in that situation will determine the final outcome for each group’s member.
36
ADDITIONAL INSTRUCTIONS FOR EMPLOYERS This experiment consists of three parts: PART 1, PART 2, and PART 3.
For each part, you will be randomly assigned to one of these roles, either Government or Manager. Your role may change during the experiment. You will be randomly and anonymously matched to new participants in each part so that you will not interact with the same persons.
In each part of the experiment, you will be asked to make a decision. Decisions that will be made in one part of the experiment will not affect decisions or earnings in the other part of the experiment. Please note that you will not be informed about the outcomes unless your group is chosen for payment.
