ArticlePDF Available

Free riding, democracy, and sacrifice in the workplace: Evidence from a real-effort experiment

December 2023
Journal of Economics & Management Strategy 34(2)

License
CC BY-NC 4.0

Authors:

Keio University

Teams are increasingly popular decision-making and work units in firms. This paper uses a novel real-effort experiment to show that (a) some teams in the workplace reduce their members' private benefits to achieve a group optimum in a social dilemma and (b) such endogenous choices by themselves enhance their work productivity (per-work-time production)-a phenomenon called the "dividend of democracy." In the experiment, worker subjects are randomly assigned to a team of three, and they then jointly solve a collaborative real-effort task under a revenue-sharing rule in their group with two other teams, while each individual worker can privately and independently shirk by playing a Tetris game. Strikingly, teams exhibit significantly higher productivity (per-work-time production) when they can decide whether to reduce the return from shirking by voting than when the policy implementation is randomly decided from above, irrespective of the policy implementation outcome. This means that democratic culture directly affects behavior. On the other hand, the workers under democracy also increase their shirking, presumably due to enhanced fatigue owing to the stronger productivity. Despite this, democracy does not decrease overall production thanks to the enhanced work productivity.

Content uploaded by Kenju Kamei

Content may be subject to copyright.

Received: 20 May 2023

Accepted: 24 November 2023

DOI: 10.1111/jems.12570

ORIGINAL ARTICLE

Free riding, democracy, and sacrifice in the workplace:

Evidence from a real‐effort experiment

Kenju Kamei

|Katy Tabero

Faculty of Economics, Keio University,

Tokyo, Japan

Department of Economic, Social and

Political Sciences, University of

Southampton, Southampton, UK

Correspondence

Kenju Kamei, Keio University, 2‐15‐45

Mita, Minato‐ku, Tokyo 108‐8345, Japan.

Email: kenju.kamei@gmail.com

Funding information

Murata Science Foundation,

Grant/Award Number: Year 2021; ESRC

North East Doctoral Training

Partnership, Grant/Award Number: ES/

P000762/1

Abstract

Teams are increasingly popular decision‐making and work units in firms. This

paper uses a novel real‐effort experiment to show that (a) some teams in the

workplace reduce their members' private benefits to achieve a group optimum

in a social dilemma and (b) such endogenous choices by themselves enhance

their work productivity (per‐work‐time production)—a phenomenon called

the “dividend of democracy.”In the experiment, worker subjects are randomly

assigned to a team of three, and they then jointly solve a collaborative real‐

effort task under a revenue‐sharing rule in their group with two other teams,

while each individual worker can privately and independently shirk by playing

a Tetris game. Strikingly, teams exhibit significantly higher productivity (per‐

work‐time production) when they can decide whether to reduce the return

from shirking by voting than when the policy implementation is randomly

decided from above, irrespective of the policy implementation outcome. This

means that democratic culture directly affects behavior. On the other hand,

the workers under democracy also increase their shirking, presumably due to

enhanced fatigue owing to the stronger productivity. Despite this, democracy

does not decrease overall production thanks to the enhanced work

productivity.

1|INTRODUCTION

Teams are increasing popular in firms as decision‐making and work units (e.g., Kamei & Tabero, 2022). However, team

decision‐making and teamwork feature a coordination problem that involves complexities relating to imperfect

information, monitoring, and agency costs (e.g., Alchian & Demsetz, 1972; Marschak & Radner, 1972). Thus,

maintaining motivation among workers is particularly difficult when teams are involved in the workplace, and their

private interests conflict with group interests (e.g., Bolton & Dewatripont, 2004)—a typical example of this is moral

hazard in groups (e.g., Alchian & Demsetz, 1972; Holmstrom, 1982). Democratic culture may help mitigate conflict

within and across the teams by not only enhancing their self‐determination and intrinsic motivation to cooperate (e.g.,

Deci & Ryan, 1985,2000), but by also providing workers with opportunities to signal their willingness to cooperate with

their peers through democratic processes (e.g., Bergh et al., 2014; Connelly et al., 2011), thereby making it easy to

achieve the group optimum. In such environments, workers may decide to collectively decrease temptations by

J Econ Manage Strat. 2023;1–21. wileyonlinelibrary.com/journal/jems

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial License, which permits use, distribution and reproduction in any

medium, provided the original work is properly cited and is not used for commercial purposes.

reducing their private gains for the sake of group interests. But how large could the effects of workplace democracy per

se on productivity potentially be? Precisely what motivates workers' sacrificial behaviors?

How to overcome moral hazard in groups is an important, active question in economics and management. A large

body of research spanning several decades has found that workers have difficulty cooperating with each other when

free‐riding incentives are sufficiently strong in a social dilemma (e.g., Ledyard, 1995; Zelmer, 2003). Specifically, prior

experimental research suggests that while some people demonstrate conditional willingness to cooperate, groups

usually cannot sustain cooperation for various reasons, for example, their cooperation behaviors are heterogeneous

(e.g., Fischbacher et al., 2001), they are easily discouraged by seeing their peers free ride (e.g., Fischbacher &

Gächter, 2010); or many tend to cooperate but by less than others (e.g., Thöni & Volk, 2018). This echoes theoretical

research that describes why moral hazard arises among workers when their effort levels are not perfectly observable

(e.g., Alchian & Demsetz, 1972; Holmstrom, 1982).

Both the theoretical and empirical literature therefore discuss that

some institutional solutions, such as corporate culture and working environments, competition (e.g., internal job

ladder, tournament), punishment and rewards, monitoring, and sorting, are required to assist collaboration and

cooperation in the workplace (see, e.g., Chaudhuri, 2011 for experimental economics literature; Prendergast, 1999 for

personnel economics). This study contributes to the large body of literature by investigating the impact of workplace

democracy, workers' behavioral reactions to a reduction in incentives to shirk, as well as the reasoning behind their

voluntary sacrificial behaviors in the workplace.

This study is the first to experimentally measure the so‐called “dividend of democracy”when the decision‐making

and work units are teams. The “dividend of democracy”refers to an effect that democracy directly has on the behavior

of those involved. The role of democratic culture on worker behavior has been actively studied in the literature in

economics and management for the last two decades (see Dal Bó, 2010 for a survey). In particular, prior experiments in

economics have shown that democracy in implementing a prosocial policy boosts cooperation in experimental games,

such as public goods or prisoner's dilemma games, as it directly affects people's own behavior and beliefs on their peers'

cooperativeness (e.g., Dal Bó et al., 2010; Kamei, 2016; Sutter et al., 2010; Tyran & Feld, 2006). Among others, Tyran

and Feld (2006) and Dal Bó et al. (2010,2019) provide methods to isolate the dividend of democracy from selection bias,

showing that the dividend of democracy is large. Scholars have recently started to study the applicability of such a

dividend of democracy in a workplace setting by using a design with real‐effort tasks, but the results surprisingly

showed that democracy per se may not have strong effects in real‐effort settings (e.g., Dal Bó et al., 2019; Kamei &

Markussen, 2023; Melizzo et al., 2014). While all prior experiments on democracy used individuals as the decision‐

making unit, the present study uses teams as the decision‐making unit of policy‐making and task‐solving for the first

time, and finds a significant dividend of democracy on work productivity (per‐work‐time production).

The policy available to workers in this study is one to reduce material incentives to shirk. Collectively sacrificing

one's benefits through fostering customs, conventions, or rules with the aim of resolving conflicting interests has been

conceptually discussed in the literature in the social sciences (such as anthropology) and biology as key features of

humans. Anecdotal evidence includes costly participation in religious groups and rituals, or recreational activities in

societies (e.g., dance and festivals), food sharing (e.g., turtle hunting by islanders for funerary rituals), holding

redistributive feasts, and attending group raids and defense (see, e.g., Hagen & Bryant, 2003; Hawkes & Bliege

Bird, 2002; Iannaccone, 1992; Sosis & Alcorta, 2003; Smith & Bliege Bird, 2000; Sosis & Bressler, 2003). The mechanism

is described as follows: sacrificing serves as a costly signal of one's own quality (e.g., Smith & Bliege Bird, 2005; Gintis

et al., 2001), thus helping to coordinate with others to cooperate and bolster a cooperative atmosphere in dilemma

situations.

2,3

Several laboratory experiments used public goods games or prisoner's dilemma games to study costly

human sacrificing tendencies with high internal validity (e.g., Aimone et al., 2013; Brekke et al., 2011; Grimm &

Mengel, 2009). The findings are that some groups (individuals) do collectively (voluntarily) decide to reduce their

private returns, thereby enhancing welfare. However, to the best of the authors' knowledge, sacrificing has not been

studied in the workplace context using a naturally occurring, real‐effort task, although recently there has been a

theoretical attempt to characterize the effects of sacrificing in the workplace (Bisetti et al., 2022).

While sacrifice has received less attention in the workplace so far, it is becoming more and more relevant due to a

surge in remote working (potentially boosting shirking) triggered by the Covid‐19 crisis and technological advances. A

broad range of examples of unobserved shirking activities and countermeasure policies are readily available in the

modern workplace. For example, cyberloafing is a typical and costly issue whereby employees covertly use their

computer or internet access for personal use during work time. The issue is especially serious when they are not in an

office. The employer may decide to introduce measures to counter employees' cyberloafing, for example, by monitoring

their use of the internet, imposing internet restriction policies and penalties for breaching them, or placing technical

KAMEI and TABERO

15309134, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jems.12570 by Cochrane Japan, Wiley Online Library on [11/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

restrictions on employees' access to certain nonwork websites.

While such policies can simply be imposed from above

by managerial staff or teams, the policies can also be enacted through decentralized decision‐making. For instance, a

factory may produce mechanical parts by assigning workers to several teams to take advantage of specialization. When

their environment is democratic and they recognize that cyberloafing undermines productivity, they may

democratically decide to enact a restriction policy across the teams, with an aim of improving the performance in

the factory if they believe that productivity impacts their material benefits, such as their wages, bonuses, or rewards.

Similar scenarios are common across various employment relationships, for example, a branch in a consulting firm, or

a sales office for products (e.g., cars). Another related example is “moonlighting”by which employees work multiple

jobs, sometimes simultaneously and/or without the permission of their main employer.

For example, an employee

may commit to working 5 days per week while secretly working for another firm to earn more by shirking the main job.

Alternatively, an employee may hold a secondary side job that takes place outside of their primary work hours, but

spend time during those hours contributing to their secondary job, such as responding to e‐mails, advertising, or

checking their website. The increase in remote working in recent years makes monitoring more difficult. Policies to

make working on the side difficult and materially unbeneficial (e.g., through using a screen‐capture tool and work‐time

tracking) may be considered if such free riding significantly undermines production in the main workplace.

This paper conducts an experiment with a novel “collaborative”real‐effort task. In the experiment, worker subjects

are randomly assigned to a team of three, and three teams constitute a group. The real‐effort task requires each team to

jointly calculate the number of 4 s in a matrix whose cells contain 1, 2, 3, or 4 s. At the onset of the experiment, each

team member is assigned a number, player 1, 2, or 3, such that they have different numbers from each other. The

matrix that player kis allocated includes only number ks while the other three numbers are blacked out. Each member

counts their assigned numbers, shares the counting outcome, and jointly calculates the final answer, on the condition

that their remuneration is based on revenue sharing in the group. To mimic the conflict between work and shirk (or

another activity) in the real workplace, each member is allowed to privately and independently play a computer game,

Tetris. A shirker can privately earn some material returns from gaming on top of psychologically enjoying Tetris. The

incentive structure is therefore similar to the so‐called stag‐hunt game (e.g., Hume, 2000 [1739]; Rousseau, 2009

[1755]): all three members must work on counting to earn a reward as a team from the collaborative task, but each

member has an incentive to deviate to gaming (thereby earning some reward privately). This setup is parallel to the

real‐world examples of modern distractions at work, such as cyberloafing and moonlighting. Before the task‐solving

phase begins, a policy that reduces the incentive to play Tetris (“reduction policy,”hereafter) is implemented in a group

either democratically (by voting) or autocratically (randomly by the computer without voting). The two treatments

(democratic, or autocratic) are designed using a between‐subjects design.

In addition to the contribution to the literature on workplace democracy, this research is novel in two additional

aspects. First, this study provides significant methodological contributions with the newly used “collaborative”

counting task and gaming as a real activity. While much research has been conducted using real‐effort tasks, a

significant issue has been reported by Araujo et al. (2016) that workers' incentive elasticity of outputs may be too small

with the real‐effort tasks used. Recently, Corgnet et al. (2015) and Kamei and Markussen (2023) allowed subjects to use,

respectively, internet browsers and comedy videos, as real leisure activities. Both of the papers showed that such

activities enhance incentive elasticity in experiments. The present paper adds to the literature by using gaming as a

real, but controlled, leisure activity for the first time in a computerized real‐effort experiment. Further, the members of

each team jointly work on a collaborative counting task. While an individual counting zeros task is widely used in the

literature (e.g., Abeler et al., 2011; Falk et al., 2006; Kamei & Markussen, 2023), the use of a collaborative version is the

first attempt in the literature, to the authors' knowledge. This design is meaningful as collaboration is a central aspect

of teamwork in many firms and organizations, and the new task is designed to explicitly simulate the coordination

structure. Notice the stark difference in the game structure between the standard counting task and the collaborative

counting task. The new collaborative one is a coordination game: individuals earn from the team task only when all

three members work by spending time counting and communicating accurately and effectively. The new task allows

researchers to study coordination in the structure of a stag‐hunt game in a natural way, even outside the research

agenda of organizational economics and management.

Second, the experiment is the first to investigate workers' sacrifice decisions and their reasoning in a real‐effort

environment. While prior research used experimental games such as public goods games to propose that some

individuals will reduce their private gains in dilemma situations, showing that such decisions may lead to a Pareto

improvement empirically (e.g., Aimone et al., 2013; Brekke et al., 2011; Grimm & Mengel, 2009), its validity in the

workplace setting is unclear as little research used naturally occurring, real effort in their experiments. Equally

KAMEI and TABERO

important is that no research explores what may drive workers to sacrifice their private gains, because no data is

available regarding their thinking. Subjects in the present experiment decide whether to reduce their private gains

through communication within their team as a team decision. This design enables us to collect a unique incentive‐

compatible data set to study the reasoning behind sacrifice decisions. A well‐established coding exercise is applied to

the communication logs to uncover reasoning effectively.

The experiment results reveal some teams' preferences for sacrifice and evidence of a dividend of democracy. 40.9%

of teams voted to reduce the incentive to play the game, and as a result, the reduction policy was enacted for 38.7% of

groups. Teams that were involved in democratic decision‐making exhibited significantly higher work productivity, that

is, performance per minute of working, than those in the regime where the computer randomly decided policy

implementation, whether the reduction policy was imposed or not. This means that the democratic culture per se

directly affected behavior. Having said that, the workers under democracy reduced work time compared to those under

autocracy, presumably due to fatigue accumulating more quickly for the former. Nevertheless, the former did not

decrease team production overall thanks to the enhanced work productivity.

The present paper also provides reasoning behind workers' sacrifice decisions based on a standard coding exercise.

It reveals that the units that planned to exclusively work on task‐solving, believed that the reduction policy would deter

others from shirking, or those that had supportive team atmospheres supported the reduction policy. It also uncovers

the value of signaling through sacrificial decisions to encourage collaboration: teams who believed that other teams

would complete tasks following the vote performed strongly.

The rest of the paper proceeds as follows: Section 2summarizes the experimental design, and Section 3reports the

results. Section 4provides insights obtained from an analysis of communication dialogues, and Section 5concludes.

2|EXPERIMENTAL DESIGN

The experiment is designed using a collaborative real‐effort task devised for this study. At the onset of the experiment,

worker subjects are randomly assigned to a team of three. The three members are then randomly assigned ID numbers,

1, 2, or 3, so that each member receives a different number from one another. Anonymity is retained such that they do

not know the identity of the other members (e.g., faces, names, gender). Let us call the player who is assigned number

k∈{1, 2, 3} “player k.”The team composition and the assigned ID numbers do not change for the entire experiment

(partner matching). Three teams further constitute a group (each group thus has nine members). The group

composition also does not change throughout. Section 2.1 explains the nature of the collaborative team real‐effort task,

after which Section 2.2 explains the structure of the experiment, a summary of treatments, the remuneration system,

and the reduction policy that could be implemented in each group. Supporting Information Appendix Asummarizes

the experimental procedure and includes instructions used in the experiment.

2.1 |A collaborative real‐effort task

Three members in a respective team collaboratively solve a variant of the counting task (“collaborative counting task”).

The original “counting task”(e.g., Abeler et al., 2011; Falk et al., 2006; Kamei & Markussen, 2023) is an individual real‐

effort task in which subjects independently count the number of 0 s in a matrix that contains 0 s and 1 s. To the authors'

knowledge, no collaborative version of the counting task has been devised and used in any prior experiments. In the

new collaborative counting task, the three team members are provided with a 15 × 15 matrix, each cell of which has a

randomly generated integer between 1 and 4 (each integer is independently drawn with a probability of 25%), and are

then asked to submit the number of 4 s. Collaboration is required to find the correct answer, because only number ks

appear on the computer screen of player k, while the other three numbers are blacked out—see Figure 1for a screen

image for player 1. Each team can find the correct answer if player kcounts the number of ks correctly and shares it

with their teammates, and the team calculates the number of 4 s accurately after that. For example, if the numbers 1, 2,

and 3 s are, respectively, 32, 14, and 43, then the correct answer (the number of 4 s) is 225 −32 −14 −43 = 136. A

calculator is available on each subject's computer screen. How to calculate the number of 4 s, and by whom, is up to

each team's discretion. When the team decides on and wants to submit the answer, all three members must submit the

team's joint answer on their own computer screens. Hence, in the submission stage as well they must communicate

with each other about their team's decision to answer correctly. In the case of disagreement, a member can submit a

KAMEI and TABERO

different answer from the others.

However, the answer will then be counted as incorrect. Once all three members

submit an answer, a new 15 × 15 matrix with randomly generated 1, 2, 3, and 4 s in each cell is assigned to the team,

and the process repeats.

Free‐form communication is available using an electronic chat window during the entire task‐solving process

(Figure 1; Supporting Information Appendix Aalso includes the screen image of the chat window), and messages are

recorded. This design piece helps the researchers study the reasoning behind members' behaviors, postexperiment.

While any sort of communication, such as discussing strategy to solve the problems, sharing the number of ks, or

chatting about unrelated matters, is allowed, subjects are prohibited from using any kind of offensive language or

sharing any information that compromises anonymity.

The more questions a team answers correctly, the higher the earnings they can generate in their group. Each correct

answer is rewarded with 180 UK pence in the experiment. How the 180 pence are distributed within the team or the

group is explained in Section 2.2.

2.2 |The experiment

There are two treatments that vary by changing the process to decide whether to enact a policy to curb members'

shirking or not. A between‐subjects design is used to avoid behavioral spillover (e.g., Bednar et al., 2012) or possible

spill‐over effects of democracy (e.g., Kamei, 2016). The experiment begins with a practice phase, which is the same for

all subjects in the experiment. The main task‐solving phase begins after the practice phase and differs by the

treatment.

The practice phase plays a role in not only familiarizing subjects with the collaborative counting task, but

also providing them with an opportunity to try the task and learn their ability to solve it.

In the practice phase, each team performs the collaborative counting task for 3 mins.

While they can answer as

many questions as they wish, they are not informed whether they answer each question correctly during the 3‐min

FIGURE 1 A screen for collaborative counting task. A screen image for player 1. The numbers 2, 3, and 4 s are blacked out on the screen

that player 1 sees. The 15 × 15 matrix in this figure is for illustration only.

KAMEI and TABERO

period. They are instead informed of the number of correct answers at the end of the practice phase. Remuneration is

based on revenue sharing in the team. This means that the money a team earns is equally divided among the three

team members (each member receives 60 = 180/3 UK pence for a correct response). Each team does not interact with

the other two teams in their group in this practice; nor are they informed of the performances of the other teams.

In the main task‐solving phase, each team performs the collaborative counting task for a much longer duration—

35 mins—with a revenue‐sharing rule in their group. This means that the credit of each correct answer (180 UK pence)

is equally shared among the three teams, that is, nine individuals as each team has three members. The marginal per‐

capita return is calculated as 20 (=180 × 1/9) UK pence.

There are two more distinct aspects in the main task‐solving phase. First, unlike the practice phase, each member

can privately shirk by playing Tetris. They can do so by simply pressing the “Game”button (Figure 2a). The screen is

then switched to the Tetris site (Figure 2b). No one, including their teammates, is made aware of a member's shirking

unless the member voluntarily reports their behavior using the electronic chat window. Further, the shirker earns a

return by staying in the Game screen: 18 pence/min spent in the Game screen.

They can return to the work site from

the Game site at any time. Workers are not allowed to work while playing Tetris, whose requirement enables the

researchers to quantify shirking versus work time as their work decisions. It should be noted here that the design of

gaming was carefully made to enhance external validity, as workers often have alternative activities available when

shirking in the workplace rather than being inactive. An advantage of using gaming over internet browsing (Corgnet

et al., 2015) as an alternative activity is the high level of control: workers may use internet browsers differently as their

preferences are heterogeneous. This feature shares similarities with Kamei and Markussen (2023) that adopted comedy

FIGURE 2 A screen image for collaborative counting task in the main task‐solving phase. (a) Work site and (b) game site.

KAMEI and TABERO

video clips as an alternative activity. However, using a game is better than video clips because implementation is

difficult with the use of the latter. While headsets were provided to each subject in Kamei and Markussen (2023), the

authors acknowledged that even a small ripple of laughter and sounds could contaminate the data. In contrast, gaming

is a purely independent, quite leisure activity.

Notice that with the gaming option, the incentive structure of the team task in the main phase is one of the so‐called

“stag‐hunt game”if they are highly skilled. Each team member can earn a small material gain with certainty by

deviating from collaboration. However, they earn a large team payoff when all three team members work on the

counting task, if each of them can count numbers sufficiently quickly.

Second, there is a penalty of three pence per incorrect answer in the main task‐solving phase. This penalty is

imposed on the team that commits the error, not the whole group. Such penalties are commonly used in the real

workplace; for example, poor performance or mistakes can result in monetary or social sanctions, increased threat of

dismissal (through escalation procedures or informal threats), or reduced pay where performance‐related wages or

team bonuses are in place (see, e.g., Doellgast & Marsden, 2019; Gibbons & Henderson, 2013; McNamara et al., 2022).

The penalty is equally shared among the three members in the team (i.e., one penny is deducted from the payoff per

team member). In short, the payoff of member iin team kcan be expressed as Equation (1):















πcicg c ic rg(, ,)=20 −+·

ki k k i

nki

(1)

where c

and

are the numbers of, respectively, correct and incorrect answers by team k,giis the time (minutes) that

member ispends in the Game screen, and ris per‐minute return from shirking. Notice that their work time is 35 −g

they are not allowed to work while playing Tetris. Using the revenue‐sharing rule per group and the alternative leisure

opportunity, the aim is to model the work environment as a tension across teams between task‐solving and gaming

(i.e., social dilemma). As intended, gaming was a privately optimal option for almost all teams in the experiment

sessions—see Section 3.3.

Worker subjects are not informed of how many questions their teams or other teams answer correctly during the

35‐min task‐solving phase.

Instead, at the end of the task‐solving phase they learn (a) the total number of correct and

FIGURE 2 (Continued)

KAMEI and TABERO

incorrect responses of their own team and (b) the total number of correct responses in their group. This setup is

realistic; for example, in manufacturing, the manager will learn how many defectives they have among mechanical

parts produced in a given day, only after quality checks at specified intervals.

At the beginning of the main task‐solving phase, the return from staying in the Game screen (r) could decrease from

18 to 16 pence/min. Notice that the size of the incentive change is very small at only two pence. This means that the

reduction policy can be thought of as a nondeterrent sanction policy, that is, a policy that does not alter the privately

optimal behaviors of workers in a group (e.g., Kamei, 2016; Tyran & Feld, 2006). As briefly reported in Section 3.3., this

interpretation turns out to be correct in the experiment: gaming was a privately optimal choice for almost all teams,

whether the reduction policy was in place or not, due to the strong incentives to free ride on other teams' work efforts.

The process to implement the reduction policy differs by treatment. There are two treatments, one with exogenous

imposition of the reduction policy (“EXO”[Exogeneous], hereafter), and one where the policy is endogenously selected

(“ENDO”[Endogenous], hereafter). In the EXO treatment, the policy is imposed in each group by the computer

randomly (i.e., with a probability of 50%). By contrast, in the ENDO treatment, the policy is implemented based on

majority voting by the three teams.

The voting procedure follows three steps:

Step 1. The three members in each team are given 3 mins to discuss, using an electronic chat window (e.g.,

Kamei, 2019b; Luhan et al., 2009), whether they want to reduce the per‐minute earnings gained by staying in the Game

screen. The communication contents are not revealed to any other team. See Supporting Information Appendix A.3 for

a screen image of this step.

Step 2. After the 3‐min discussion, the three members each submit their preferred decisions. If the three submit the

same decision, it becomes their team vote. However, in the case of disagreement they can submit whatever they prefer,

in which case whichever receives at least two members' support is implemented as their team vote.

Step 3. The reduction policy is implemented in the group based on majority voting. Specifically, it is implemented

(not implemented) if it receives two or three supporting (opposing) team votes. All subjects in the group are informed

of the vote outcome and the number of supporting votes.

Notice that as the reduction policy, despite the size of the reduction being small, may encourage teams to work

harder through decreasing the material incentives to shirk, thereby leading to a higher payoff, groups may decide to

decrease such private returns by voting. As summarized in Table 1, there are four possible institutional outcomes in

this study.

2.3 |Theoretical predictions

Theoretical predictions on the dividend of democracy can be derived by setting a utility function for the player and then

finding their utility‐maximizing behavior. As shown in online Supporting Information Appendix B, a calculation

TABLE 1 Treatments, Distribution of Votes and Institutional Outcomes.

Treatment and

institutional outcome

Condition in which the policy

is/is not implemented # Subjects

# Of subjects in

proreduction teams

# Of subjects in

antireduction teams

ENDO treatment Voting 279 114 165

(i) Policy was implemented At least two teams vote for

the policy

108 75 33

(ii) Policy was not

implemented

At least two teams vote

against the policy

171 39 132

EXO treatment By the computer 273 ––

(i) Policy was implemented Randomly (50% probability) 123 ––

(ii) Policy was not

implemented

Randomly (50% probability) 150 ––

Total –552 114 165

Note: The numbers in the “# of subjects in proreduction teams”and “# of subjects in antireduction teams”columns are based on the results of voting in the

experiment.

KAMEI and TABERO

suggests that teams work harder with than without the reduction policy in a given institutional condition (ENDO or

EXO), and that the positive effect is stronger in the ENDO than in the EXO treatment, for the following reasons. First,

the positive effect of the reduction policy holds theoretically for the EXO treatment because the policy reduces the

material incentives of shirking. As the reduction policy is imposed randomly in each group, in theory there are no

differences in individual characteristics between the groups where the policy is imposed or not. Thus, only the material

incentives matter in this treatment due to the lack of selection. Second, the positive effect is also applicable to the

ENDO treatment, not only due to the beneficial effects of incentive changes, but also possible selection effects through

voting. The reduction policy is enacted in the ENDO treatment only when the majority of teams support the policy.

Considering that teams who are better at solving the collaborative counting task can be assumed to incur smaller effort

costs for a given effort level, the beneficial effects of the policy on hard work exceed enhanced effort costs more easily

for such higher‐skilled teams. This means that higher‐skilled teams are more likely to enact the reduction policy by

voting, and to perform strongly in the ENDO treatment. In other words, the impact of the reduction policy is detected

more strongly in the ENDO treatment due to selection.

It should be worth remarking here that, theoretically, the positive effect of the reduction policy does not emerge

when task‐solving is too costly for teams. If the return from shirking as a team is much larger than the marginal return

from working, members in selfish teams will just stay in the Game screen even when the sorting effects are present in

the ENDO treatment.

The main hypothesis of the paper is on the dividend of democracy summarized below:

Hypothesis. Teams put more effort into task‐solving in the ENDO than in the EXO treatment,even after

controlling for possible selection effects.

The phenomenon summarized in this hypothesis is the so‐called dividend of democracy. Its mechanism lies in the

democratic process that directly influences worker tendency (e.g., Dal Bó et al., 2010,2019; Kamei, 2016; Sutter

et al., 2010; Tyran & Feld, 2006). In a workplace setting, Kamei and Markussen (2023) model this effect such that

workplace democracy lowers workers' marginal effort costs. A model similar to Kamei and Markussen (2023) supports

the hypothesis above; a decrease in the marginal effort costs driven by democracy results in hard work among teams

(see Supporting Information Appendix Bfor the detail). Part of the dividend of democracy can also be attributed to

signaling effects (e.g., Jensen & Markussen, 2022; Kamei, 2019a; Tyran & Feld, 2006).

It should be noted that identifying the dividend of democracy requires care because of the possible selection bias

already discussed (Dal Bó et al., 2010,2019; Tyran & Feld, 2006). By design, proreduction teams are overrepresented

(underrepresented) in groups where the reduction policy was (was not) endogenously enacted. As voting behavior is

likely related to teams' skills and work behavior, group behaviors are not comparable between the ENDO and EXO

treatments unless the distributions of votes are balanced. The present paper adopts the “weights‐based identification

strategy”proposed by Dal Bó et al. (2019). This estimation method uses weights under the whole population when

calculating the average behavior in the ENDO treatment, rather than the realized vote shares in specific institutional

outcomes. For instance, suppose that 50% of teams vote for the reduction policy and the policy is imposed in 50% of

groups. The % of proreduction teams would be much more (less) than 50% in groups where the policy is (is not)

endogenously imposed because of majority voting. Instead of the high (low) percentage in such groups, 50% is used as a

weight in calculating the average behaviors of pro‐and antipolicy units with this method. The detail of the reweighting

method along with the data will be provided in Section 3.

3|POLICY PREFERENCES, AND THE DIVIDEND OF DEMOCRACY

A total of 552 students (279 for the ENDO treatment and 273 for the EXO treatment) at the University of York in the

United Kingdom participated in the experiment. No subjects participated in more than one session. The experiment

followed standard practices in economics, such as neutral framing. Supporting Information Appendix Aincludes the

procedure and the instructions.

Table 1of Section 2includes the distribution of team votes in the experiment. Consistent with the literature on

voting experiments among individuals (e.g., Aimone et al., 2013; Dal Bó et al., 2010), it reveals that some teams vote to

reduce their private returns from shirking. It indicates that 40.9% of teams (=38/93 × 100%) voted for the reduction

policy. As a result of majority voting, the policy was enacted in 38.7% (=36/93 × 100%) of groups in the ENDO

KAMEI and TABERO

treatment. Table 1also shows a clear pattern of selection bias. In the ENDO treatment, the percentage of proreduction

teams was 69.4% (=25/36 × 100%) in groups where the policy was enacted, while the percentage was only 22.8% (=13/

57 × 100%) in groups where it was not enacted. Hence, proreduction teams were overrepresented (underrepresented) in

groups where the reduction policy was (was not) enacted in the ENDO treatment. This is a pattern similar to the

selection bias discussed in Dal Bó et al. (2010,2019).

In fact, teams' support for the policy was positively correlated with their performance before voting. In the practice

phase, teams performed the task for only 3 mins under individual‐based remuneration. The data indicate that teams

which voted for the reduction policy on average answered 1.001 questions correctly in the practice phase; their

performance was significantly better at two‐sided p< .01 (z= 4.230) than teams which voted against the policy (the

average number of correct answers by antireduction teams was 0.414). This pattern holds regardless of the institutional

outcome, that is, whether the policy was enacted or not (Supporting Information Figure C.1). This means that

proreduction teams may have characteristics different from antireduction teams. As shown in Supporting Information

Figure C.1, the performance of teams in the EXO treatment was somewhere in the middle of the pro‐and antireduction

teams (was similar to that of antireduction teams) in groups where the policy was enacted (was not enacted).

In sum, selection bias must be controlled for when identifying the dividend of democracy in the data. This paper

utilizes the method proposed by Dal Bó et al. (2019) to remove selection effects. Section 3first discusses the dividend of

democracy on work productivity, after which it discusses workers' effort choices in detail and their welfare

consequences.

Result 1. 40.9% of teams voted for reducing returns from staying in the Game screen. As a result of majority voting,

the reduction policy was enacted in 38.7% of groups in the ENDO treatment.

3.1 |Dividend of democracy on work productivity

The first key result of this study is the positive effect of democracy on work productivity. The dividend of democracy is

quite strong: around 20% on average. Consider, first, groups where the reduction policy was enacted. Productivity,

defined as the number of correct answers per minute of teamwork (i.e., per average time spent in the task screen by a

team member), is 0.594 in the ENDO treatment. The 0.594 means that if a team, that is, all three members, worked the

entire 35 mins of the task‐solving phase without playing Tetris, they would be able to answer on average 20.79

(=0.594 × 35) tasks correctly. This productivity is 28.5% larger than the productivity in the EXO treatment, which is

calculated as 0.462.

Part of the productivity increase can be attributed to selection bias as already discussed. Thus,

such bias must be controlled for to isolate the dividend of democracy by adjusting the “weights,”that is, the

distribution of votes. This paper follows Dal Bó et al. (2019) calculating the reweighted productivity with the following

two steps:

Step 1. Calculate (a) the average number of correct answers and (b) the per member average work time, using as

weights the percentage of proreduction teams in the population (40.9%) rather than the percentages under the

reduction regime in the ENDO treatment (69.4%).

Step 2. Calculate (a)/(b).

The reweighted work productivity in the ENDO treatment found using these steps is still quite large—that is, 0.529,

14.5% larger than that in the EXO treatment.

Consider, next, groups where the reduction policy was not enacted. There is also a strong effect of democracy for

these groups. First, the productivity before reweighting was modestly different between the two conditions: 0.488 in the

ENDO and 0.431 in the EXO treatment. However, this mild difference is due to selection, in that proreduction teams

are underrepresented in the ENDO treatment, that is, these account for only 22.8% of teams (Table 1). Productivity after

reweighting was large, 0.539, in the ENDO treatment. This means that the dividend of democracy is 0.108

(=0.539–0.431) correct answers per min. of teamwork, that is, a 25.1% increase in productivity. The fact that democracy

strongly affects behavior irrespective of the policy implementation outcome suggests that being involved in the

democratic process by itself, that is, democratic culture, affects their work motivation directly, which is consistent with

the idea that democracy directly enhances intrinsic motivations to work (e.g., Deci & Ryan, 1985,2000).

In sum, the reweighted dividend of democracy without the reduction policy (i.e., 0.539 vs. 0.431) was of almost a

similar magnitude to the one in groups with the reduction policy (0.529 vs. 0.462). This underscores the strong role of

democracy in improving productivity. For this reason, the two institutional outcomes (with or without the policy) are

pooled to statistically test the significance of the dividend of democracy (Table 2).

KAMEI and TABERO

Table 2reports test results for the dividend of democracy on work productivity using all of the data. To calculate

each pvalue, the estimates for the dividends of democracy were calculated 20,000 times based on session‐level

bootstrapping.

Panel A of Figure 3reports the distributions of estimated dividends of democracy. These reveal that

the size and the significance of the dividend of democracy are only slightly affected by the correction of the selection

bias. The overall impact is economically large: democracy boosts productivity by 20.02% (=(0.535 −0.445)/

0.445 × 100%) and it is significant at the 5% level. Hence, it can be concluded that democracy by itself strongly

improves productivity.

Readers may also be interested in knowing how the dividend of democracy persists in the workplace. To answer

this question, work productivity measures are calculated by splitting the data into quarters of the experiment. It first

shows that experience does help improve teams' problem‐solving skills, and hence their per‐minute‐of‐teamwork

performance. Panel B of Figure 3indicates that, whether in the ENDO or EXO treatment, work productivity increased

from quarter to quarter. The dynamics also reveal that higher work productivity in the ENDO treatment, relative to

EXO treatment, was remarkably stable throughout the experiment. This means that fatigue (whether physical or

mental) and/or monotony may not weaken the dividend of democracy in the workplace.

Result 2. (a) There is strong evidence that democracy significantly boosted work productivity, defined as the

production per minute spent working. (b) The positive dividend of democracy persisted throughout the task‐solving

phase.

While the strong role of democracy is consistent with the findings from prior research on democracy using

“experimental games,”such as prisoner's dilemma and public goods games (e.g., Dal Bó et al., 2010; Kamei, 2016;

Tyran & Feld, 2006; Sutter et al., 2010), it is at odds with the finding from the “real‐effort”experiment of Kamei and

Markussen (2023). In Kamei and Markussen (2023), subjects were assigned to a group of three and then worked

individually on either the “counting task”(e.g., Abeler et al., 2011; Falk et al., 2006) or the “addition task”(e.g., Corgnet

et al., 2015; Niederle & Vesterlund, 2007) on condition that a revenue‐sharing rule is in use and a funny video is

available as an alternative activity. Kamei and Markussen (2023) found little evidence of the effects of democratic task

selection. The null result was indeed a puzzle which Kamei and Markussen (2023) were not able to explain. A similar

null result for the dividend of democracy was also observed and posed as a puzzle in the real‐effort experiment of Dal

Bó et al. (2019) where internet surfing (e.g., Corgnet et al., 2015) was available as an alternative activity. So, why did we

get a strong dividend of democracy in the present study? A likely reason is that each team member had stronger

shirking opportunities in the present study. Subjects in the present experiment jointly solved a collaborative counting

task as a team in a group, unlike in the prior experiments where subjects individually solved an individual real‐effort

task in a group. Specifically, while incentives to shirk as a decision‐making unit (teams in this study or individuals in

the other research) in a group are the same, each team member in the present study has additional opportunities to

shirk by playing Tetris privately, that is, without notifying their other team members, whose structure features a

coordination game inside the team.

The difference between the present and the earlier experiments suggests that the

dividend of democracy may be more important in an environment where workers have stronger incentives to shirk.

TABLE 2 Dividend of democracy in work productivity.

(A) Using original weights (B) Using adjusted weights following Dal Bó et al.

Team production per minute of its three members' working

(a) ENDO treatment 0.536 0.535

(b) EXO treatment 0.445 0.445

Two‐sided pfor H

: (a) = (b)

0.036** 0.046**

Notes: The overall productivity measures in rows a and b were calculated using the distribution of policy implementation in the EXO treatment (i.e., % of

groups with policy: % of groups without policy = 123/273 : 150/273). The numbers in column (A) are productivity measures calculated using the original

distributions of voter types under institutional outcomes (pro‐or antireduction teams) shown in rows (i) and (ii) under the ENDO treatment of Table 1. The

numbers in column (B) are productivity measures using the distribution of voter types in the population following the weights‐based identification strategy

proposed by Dal Bó et al. (2019).

The number of correct answers per minute of teamwork

The pvalues were calculated using the bootstrapping procedure described in Dal Bó et al. (2019). The number of bootstrap iterations was 20,000 (Figure 3).

**p< .05.

KAMEI and TABERO

3.2 |Effort choices and welfare

The larger size of work productivity (Result 2) does not mean that democracy improves production in the workplace.

Rows (I) and (II) of Table 3report the average numbers of attempts and correct answers in the main task‐solving phase.

The average results are reported by the policy implementation outcome because work behaviors differed substantially

by the presence of the reduction policy. It shows that teams attempted more questions and, as a result, answered more

questions correctly, in the ENDO than in the EXO treatment (rows I and II). However, the positive effects of democracy

are far from significant (columns 2, 2a, and 2b).

This insignificant impact, despite Result 2, was due to the workers' effort choices. As the collaborative counting task

was a relatively challenging real‐effort task, shirking prevailed in the experiment.

Workers (although insignificantly)

(a)

(b)

FIGURE 3 Dividends of democracy for work productivity. (a) Distribution of bootstrapped dividends of democracy for productivity

based on Dal Bó et al. (2019). (b) Dividend of democracy, quarter by quarter. (1) Each distribution in panel A was drawn using 20,000

estimated dividends of democracy based on bootstrap iterations. (2) The productivity measures of each quarter in panel B were calculated by

splitting the duration of the 35 task‐solving phase by four (e.g., the first quarter is the first 35/4 mins).

KAMEI and TABERO

shirked more on average in the ENDO treatment than in the EXO treatment—see columns (2), (2a), and (2b) of row

(III). The higher incidence of shirking undermined the positive impact of enhanced work productivity, which resulted

in the insignificant effect on the two effort output measures. Thus, this result suggests that a firm needs to have some

mechanism to curb workers' effort choices beyond democracy if they want to increase production significantly, because

workers' discretion to decide how much to work may partly cancel out the sustained positive dividend of democracy.

However, it should be emphasized here that despite the increased shirking, team production did not decrease (instead

increased although insignificantly) thanks to Result 2 under democracy. This means that the same level of production

can be achieved under democracy in less work time.

One may wonder why democracy worsened shirking. One possible interpretation here is that democracy

enlarged workers' motivations to earn a high payoff in the experiment. The subjects may have perceived it to be

more payoff‐enhancing if they worked harder for a shorter duration and then secured certain gains from staying on

the Game screen once exhausted. Although it cannot be verified, this possibility may partly explain the behavior

since, despite Result 2(b), subjects may quickly have feelings of fatigue if their per‐minute effort levels rise. Having

said that, such a reduction in work time did not work well for the workers, since, while democracy did increase the

average payoff, the impact is insignificant after controlling for selection (row IV). This implies that their effort

choices were not optimal. But, if this conjecture is relevant, why did perceived fatigue play a large part in the

behavioral decisions of experiment subjects? A likely possibility is that Result 2 was still not enough to encourage

workers to choose putting in a greater effort over shirking. This possibility is quite reasonable as discussed

carefully in Section 3.3.

Result 3. Despite Result 2, democracy did not increase team production significantly, because workers under

democracy decreased work time to some degree.

TABLE 3 Work performance and the dividend of democracy.

Unweighted Reweighted

All data With policy Without policy All data With policy Without policy

(1) (1a) (1b) (2) (2a) (2b)

(I) Average number of attempts

(a) ENDO 19.49 25.28 14.74 18.81 20.53 17.40

(b) EXO 16.79 19.49 14.58 16.79 19.49 14.58

: (a) = (b) 0.151 0.043** 0.949 0.331 0.747 0.285

(II) Average number of correct answers

(a) ENDO 12.49 16.61 9.12 11.96 13.14 11.00

(b) EXO 10.49 12.12 9.16 10.49 12.12 9.16

: (a) = (b) 0.170 0.060*0.983 0.330 0.671 0.336

(III) Average per member time spent in the Game screen (minute)

(a) ENDO 12.14 7.05 16.31 12.60 10.14 14.61

(b) EXO 11.50 8.79 13.72 11.50 8.79 13.72

: (a) = (b) 0.664 0.345 0.236 0.534 0.594 0.711

(IV) Average payoff in the main task‐solving phase (pound sterling)

(a) ENDO 9.62 11.09 8.41 9.35 9.51 9.23

(b) EXO 8.29 8.68 7.97 8.29 8.68 7.97

: (a) = (b) 0.065*0.062*0.555 0.138 0.498 0.150

Notes: The pvalues were calculated using the bootstrapping procedure described in Dal Bó et al. (2019). The number of bootstrap iterations was 20,000. The

numbers in columns (1), (1a), and (1b) were calculated using the original distributions of voter types under institutional outcomes (pro‐or antireduction

teams) shown in rows (i) and (ii) of Table 1. The numbers in columns (2), (2a), and (2b) were calculated using the distribution of voter types in the population

following the weights‐based identification strategy developed by Dal Bó et al. (2019). The overall measures in columns (1) and (2) were calculated using the

distribution of policy implementation in the EXO treatment (i.e., % of groups with policy: % of groups without policy = 123/273 :150/273).

*p< .1; **p< .05.

KAMEI and TABERO

3.3 |Privately versus socially optimal behaviors

This experiment was designed to model a social dilemma problem, that is, conflicts among teams, in the

workplace. Section 3.3 briefly checks the validity of this design setup, finding that its attempt was successful as

intended. This section also tries to find an answer as to why democracy was not enough to boost team production

in the experiment.

Since staying in the Game screen was remunerated with 16 or 18 pence/min, it is possible to calculate for what

percentage of teams task‐solving was a socially or privately optimal strategy (in the sense of material payoff

maximization). In order for task‐solving to be privately optimal, a team needs to be able to solve at least 0.80 = 16/20

(0.90 = 18/20) tasks correctly per minute when the reduction policy is (is not) in place. A detailed look at the data

(Supporting Information Appendix Table C1) indicates that gaming was a privately optimal choice for almost all teams

in the EXO treatment, whether the policy was in place or not. Specifically, it is so for 95.60% of teams (87 out of 91

teams) in the EXO treatment.

This implies that the reduction policy was nondeterrent in the experiment. Consistent

with the prior experimental evidence on exogenously introduced nondeterrent punishment (e.g., Kamei, 2016; Tyran &

Feld, 2006), the effect of the reduction policy was not large in the EXO treatment. Specifically, while the average

number of correct answers in the EXO treatment was larger with than without the reduction policy (12.12 vs. 9.16), the

difference was not significant at two‐sided p= .109 according to the bootstrap method used in the other tests of the

paper (the difference is significant but only at the 10% level, that is, p= .0707 if a two‐sided Mann–Whitney test is

used).

However, as intended, the socially optimal strategy was task‐solving for many teams. In order for task‐solving to be

socially optimal, a team needs to be able to solve at least 0.227 ≈16/60 (0.30 = 18/60) tasks correctly per minute when

the reduction policy is (is not) in place. Overall, the social optimal condition was met for 61.6% of teams (56 out of 91

teams) in the EXO treatment. Notice that task‐solving is never privately optimal for teams whose task‐solving is not

socially optimal. Consistent with this incentive pattern, teams whose task‐solving was not socially optimal spent

significantly less time working on the task than the other teams at two‐sided p< .001 (15.29 vs. 28.63 mins in the EXO

treatment). The average number of correct answers per minute of working by the former was only 0.07, but that by the

latter was 0.54 in the EXO treatment.

In sum, the present experiment can be thought of as exploring workers' voting and effort choice decisions under

social dilemmas in the workplace when the target was a nondeterrent reduction policy.

Then, one may ask whether democracy might have altered the social dilemma situation to another one (e.g.,

coordination game), as arguably democracy not only enhances work productivity (Section 3.1), but also reduces effort

costs in task‐solving. Another look at the data, however, shows that the answer is negative. Specifically, a calculation

finds that gaming was a privately optimal choice for almost all teams in the ENDO treatment, that is, 91.40% of teams

(85 out of 93 teams); and task‐solving was a socially optimal choice for 61.3% (57 out of 93 teams) in that treatment—

see again Supporting Information Appendix Table C1. These numbers are quite similar to those in the EXO treatment

already discussed.

The reason why worker behavior was characterized by Results 2 and 3 is explained by the theoretical analysis

summarized in Supporting Information Appendix B. The model there assumes that, following the prior research

findings, democracy eases a worker's effort cost, and it also boosts their productivity (its positive effect on work

productivity is a parameter μin that model). μ> 0 was confirmed by the experiment data as summarized in Result 2.

The team's optimal effort provision can then be determined by the relative strength between (a) work productivity

(s+μin the theoretical model, where sis the marginal return of effort provision by team i) and (b) the material

incentives to shirk by staying in the Game screen. Theoretically, the positive value of μ(Result 2) possibly changes the

materially beneficial choice from gaming to task‐solving—see Supporting Information Appendix Figure B.2. However,

the analysis in the Supporting Information Appendix indicates that if the impact on work productivity is not

economically large enough, gaming is still the most materially beneficial activity even when teams have a statistically

significant dividend of democracy. This is exactly what the above calculations on privately versus socially optimal

choices in the experiment data demonstrate. The calculations clearly reveal that democracy did not change the

underlying private incentives in the experiment. This means that additional mechanisms on top of democracy would be

required to change the incentive structure so that task‐solving becomes a privately optimal choice for workers, if the

group wants to increase production.

KAMEI and TABERO

4|UNDERSTANDING SACRIFICE BEHAVIOR: COMMUNICATION

CONTENTS

The present experiment provides a useful opportunity to explore workers' reasoning behind their decisions to reduce

private returns from shirking as communication contents are available. While the decision data not only uncovered

some subjects' preferences to reduce their private returns but also detected a significant dividend of democracy on work

productivity (Section 3), it is still unclear what drove such behavioral patterns.

Two independent coders were hired to read and classify the communication contents based on their judgment of

the subjects' motives. Specifically, a list of codes was designed by the authors, based on the theoretical predictions of the

setup and related literature, that could potentially reflect a subject or teams reasoning and/or behavior. The list was

given to the coders to assign whichever codes (including none) they deemed relevant to a given communication log.

The coding procedure follows Kamei and Tabero (2022) which utilized the standard coding approach in economics to

analyze teams' behavioral reasoning in the context of institutional choices based on intrateam communication logs.

The detail of the coding procedure and the full lists of codes used for the present paper can be found in Supporting

Information Appendix Sections D.1 and D.2.

The agreement rates and Cohen's Kappa values (Cohen, 1960) can be used to judge the consistency of the coding

process between the two coders. Overall, the agreement rates (Kappa values) between the two coders were 96.9% (0.87)

and 94.8% (0.78) in the ENDO and EXO treatments, respectively. The Kappa values are at least 0.4 for 92.5% and 78.0%

of individual codes in the ENDO and EXO treatments, respectively (Supporting Information Appendix Section D.3). As

a Kappa value of 0.4 is usually used as a threshold for a researcher to judge the reliability of coding, we use only the

codes that exceed this boundary in this analysis.

Table 4summarizes the list of codes that are found to have impacted the units' voting significantly at least at the

10% level. Their voting is clearly linked to their intention regarding what to do during the main task‐solving phase

(Code Bs): while units supported the reduction policy if they planned to focus on task‐solving, they opposed it if they

were considering using the game screen. The coding category linked to pro/antipolicy reasoning (Code Cs) reveals clear

motives behind the policy preferences. While the policy is nondeterrent, those who voted in favor of it did so to deter

others from shirking (Code C1). On the other hand, those who intended to game or believed that the policy was too

weak to alter shirking opposed its enaction. Lastly, unsurprisingly, their views on materially beneficial behavior and

team atmosphere influenced voting. Specifically, units that believed their privately optimal behavior was task‐solving

supported reducing the return from gaming. By contrast, units who experienced discomfort or poor performance from

task‐solving in the practice phase opposed such a reduction. While teams with a positive atmosphere (E2) supported

the reduction policy, those with poor or lacking communication opposed it (E5).

TABLE 4 Significant code meanings and its impact on voting for the reduction policy.

Code Meaning Direction

B1 Agree/imply to count as the primary behavior (+)***

B2 Agree/imply to game as the primary behavior (−)*

B3 Agree to hybrid behavior, for example, so many tasks/minutes before switching to the game screen (−)**

B4 Agree to discuss, decide, and/or alter behavior during the counting task later (35‐min phase) based on

performance/needs in Phase 2

(−)**

C1 Propolicy to deter others from switching to the game screen by reducing the return (monetary deterrence) (+)***

C8 Antipolicy as they intend to game for at least some of the task‐solving period (−)***

C11 Express that the policy is not strong enough to deter others from switching to the game screen (monetary) (−)**

D2 Believe they as a team make the most money from counting (+)***

D5 Discuss their performance or comfort in Phase 1 (weak/negative) (−)**

E2 Positivity towards teammates, for example, attempts to encourage others or being supportive (+)*

E5 No communication from just one or two team members (−)***

Notes:+(−)In“Direction”means the reasoning en(dis)courages voting for the policy.

*p< .1; **p< .05; ***p< .01.

KAMEI and TABERO

Result 4. The units that planned to exclusively work on task‐solving, believed that the reduction policy would deter

others from shirking, or those that had supportive team atmospheres, voted for the reduction policy. However, those

who previously experienced discomfort or poor performance from working, considered (even only potentially) using

the Game screen, believed that the policy was too weak to alter peers' shirking, or had poor communication with their

teammates, voted against the reduction policy.

As summarized in Result 4, units' commitment to task‐solving and their intention to affect others' shirking were the

drivers behind their votes in favor of the reduction policy. To explore how policy implementation outcomes affected

units' behaviors, coding analyses were further performed using the communication logs of the 35‐min task‐solving

phases (Table 5). Three similar tendencies were observed for both the ENDO and EXO treatments. First, those who

TABLE 5 Reasoning behind work choice and productivity.

Code Meaning Direction

Codes related to reactions to vote outcome in ENDO (Code Fs) or policy outcome in EXO (Code Is)

(ENDO treatment)

F1 Express negative emotions (e.g., upset and anger) about the outcome of the vote (wt−)***

F3 Agree/imply to count as the primary behavior (wt+)***, (p+)***

F4 Agree/imply to game as the primary behavior (wt−)***,(p−)***

F5 Agree to hybrid behavior, for example, so many tasks/minutes before switching to the

game screen

(wt−)**,(p−)*

F6 Agree to discuss, decide, and/or alter behavior during the counting task later (35‐min phase)

based on performance/needs in Phase 2

(p−)*

F7 Express belief/hope that other teams will complete tasks following the vote (wt+)***, (p+)**

F8 Express belief that teams will not complete tasks following the vote (wt+)***, (p+)***

F9 Discuss the distribution of votes and predict how each team may respond to one another (wt+)***

F13 Belief on other teams' responses: antipolicy teams will work little (p−)***

F15 Discuss whether to change behavior based on the vote outcome (wt+)***

(EXO treatment)

I1 Express negative emotions (e.g., upset and anger) about the policy outcome (wt−)***,(p−)***

I4 Agree/imply to game as the primary behavior (wt−)***,(p−)***

I5 Agree to hybrid behavior, for example, so many tasks/minutes before switching to the

game screen

(wt−)***,(p−)**

I6 Agree to discuss, decide, and/or alter behavior during the counting task later (35‐min phase)

based on performance/needs in Phase 2

(wt−)*

Other codes (the same codes were used for the ENDO and EXO treatments)

G4 Expression of strong negative emotion, for example, frustration, anger, disappointment (p−***, Exo)

G5 Expression of strong positive emotion, for example, enjoyment, things are going well (wt+***,p+**, Exo)

D4 Discuss their performance or comfort in Phase 1 and/or so far in Phase 2 (strong/positive) (wt+***,p+***, Endo)

D5 Discuss their performance or comfort in Phase 1 and/or so far in Phase 2 (weak/negative) (wt−***,p−***, Endo),

(wt−**, Exo)

D8 Discuss uncertainty surrounding other teams' work choices or abilities (wt−***,p−**, Exo)

D9 Suggest distrust of other teams, for example, expect them to take advantage (wt−***,p−***, Endo)

Notes: wt and p in the “Direction”column indicate two work performance measures: work time (minutes) and productivity defined as the number of correct

answers divided by the work time. + (−) means the reasoning in(de)creases the performance measures. All significant codes are listed for Codes Fs and Is,

while only some key codes are included for the other coding categories to conserve space (Supporting Information Appendix D4 includes the full estimation

results).

*p< .1; **p< .05; ***p< .01.

KAMEI and TABERO

reacted negatively to the implementation outcome tended to work less (F1, I1). Such negative reciprocal tendencies

were unsurprising considering the large findings of other‐regarding preferences—see, for example, Sobel (2005) and

Fehr and Schmidt (2006). Second, a unit's plan to work on counting or engage in gaming affects performance (F3, F4,

F5, F6, I4, I5, I6), similar to Result 4. Third, units' positive and negative experiences of task‐solving, respectively,

improve and hurt performance (G4, G5, D4, D5).

The results reveal signaling effects of voting on task‐solving, and some nuanced evidence about the teams' dividends

of democracy seen in Result 2. First, units that considered the distribution of votes to predict others' task‐solving or

discussed changing behavior worked longer (F9, F15). Second, units who believed that other teams would complete

tasks following the vote performed strongly (F7), resonating with the idea that voting has a signaling value, thereby

encouraging collaboration. Further, even units who thought that others would not respond to the reduction policy

improved their performance (F8), which implies democracy directly affects behavior beyond signaling. Nevertheless, its

effects are canceled out if an antipolicy team is present in a group and units have a negative view of the task‐solving

behavior of the antipolicy team (F13).

5|CONCLUSION

Teams are popular decision‐making and work units in organizations that feature a complex coordination problem.

Overcoming moral hazard among teams in the workplace plays a crucial role in maintaining productivity in the firm,

whether in the traditional work environment or in a remote working setting, such as that triggered for many by the

Covid‐19 crisis. The present paper investigated how frequently groups reduce the return from shirking by enacting a

formal nondeterrent sanction policy, and how such endogenous choices per se improve work productivity. To achieve

this, a novel real‐effort experiment was designed, equipped with (a) a collaborative counting task featuring an

intrateam coordination game and (b) gaming (Tetris) as a real leisure activity. The experiment results showed that

around 40% of teams voted to reduce the return from staying on the Game screen. A contents analysis using teams'

communication logs showed that such voting was driven by not only their commitment to work on counting but also

their belief that the reduction policy would deter others from shirking.

The decision data uncovered a significant and strong dividend of democracy on work productivity. Strikingly,

whether the policy was enacted or not, teams in the ENDO treatment displayed significantly higher per‐work‐time

production than those in the EXO treatment. This means that democratic culture directly affects behavior positively.

However, the workers under democracy also experienced higher levels of shirking, that is, the time spent on the Game

screen was larger in the ENDO than in the EXO treatment, presumably driven by their enhanced fatigue due to the

more intensive working in the former. This implies that while additional mechanisms that affect incentives besides

democracy may be required to increase production, democracy may improve efficiency. What kinds of mechanisms

would work best to instead increase production further remains for future research. Having said that, it should be

emphasized here that the average production of the workers under democracy did not decrease (it increased, although

insignificantly) thanks to their strong per‐work‐time production.

The findings on the positive dividend of democracy on work productivity have a policy implication for effective

human resource and management practices. While prior research suggests that innovative human resource

management involving worker participation (such as that in production sites) leads to better work performance

(e.g., Ichniowski et al., 1997), it is unclear how democracy affects behavior, as earlier real‐effort experiments failed to

detect strong dividends of democracy (e.g., Dal Bó et al., 2019; Kamei & Markussen, 2023). Using an environment with

strong shirking incentives, the present experiment suggests that organizations with a shared goal can benefit from

introducing participatory decision‐making with their employees or group members, by potentially improving their

work productivity. Even when democracy induces the workers to work less, the improvement in productivity allows for

achieving a production goal with fewer working hours.

The effect of democratic culture in achieving the same goal in less work time collaborates with recent work style

reform. There is a trend to transform the traditional workplace into an employee‐centered workplace in many

countries. For example, in the United Kingdom, some firms recently tested 4‐day work weeks to make working

conditions flexible to meet the different needs of employees.

Having higher work productivity in a democratic

environment certainly helps firms achieve the same or potentially better outcomes with fewer working hours. This

boost to productivity is achieved through enhanced self‐determination and signaling effects; workplace democracy

provides the workers with the ability to foster trust with each other and to indicate their intentions or desire to

KAMEI and TABERO

cooperate through democratic procedures such as voting, and the recipients can then respond to these signals. Such a

social exchange may be fundamental for workers to achieve collaboration by reducing uncertainty surrounding each

other's behavior in a democratic workplace environment. The firm may create a positive and collaborative atmosphere

and improve productivity by designing democratic systems in multiple layers and activities across the organization

(Smith & Bliege Bird, 2000; Smith & Bliege Bird, 2005).

ACKNOWLEDGMENTS

This project was supported by a grant‐in‐aid from the Murata Science Foundation. Durham University and the

Northern Ireland and North East Doctoral Training Partnership (ES/P000762/1) provided additional funding support.

The authors thank John Hey and Mark Wilson (an IT manager at the University of York) for their support in recruiting

subjects, and Louis Putterman and Pedro Dal Bó for helpful comments.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon request.

ORCID

Kenju Kamei http://orcid.org/0000-0002-7905-8584

ENDNOTES

The difficulty in sustaining cooperation has also been widely discussed in the theoretical literature on the voluntary provision of public

goods (e.g., Bergstrom et al., 1986; Samuelson, 1954).

In general, actors' many decisions are characterized as costly signaling in modern societies. Examples include the job market, in which

applicants invest in education or other qualifications to indicate their quality (Spence, 1973), or at the firm level by which firms indicate

their quality to other firms, the market, or other stakeholders through investment in high profile board members, awards, alliances, or

underpricing (see Bergh et al., 2014 for a review and examples).

Empirically, people are known to choose transaction partners in dilemma situations based on factors that inform the quality of that

partner. Elfenbein et al. (2012), using a novel data set composed of more than 160,000 eBay listings, successfully demonstrated that in

online marketplaces, buyers tend to purchase products tied to charity, and thus sellers have incentives to use a charity program (e.g.,

eBay's Giving Works program) as a quality signal.

Bisetti et al. (2022) propose a self‐reporting mechanism in which a team's pay is based on their observed joint output and their team's

self‐reported performance. They prove that a team has the incentive to under‐report their group's performance (sacrifice wages for all in

the team) as a punishment to free‐riders, thereby enabling them to achieve higher welfare.

Strengthening monitoring increases the probability that cyberloafing is detected and penalties are assigned, thereby reducing workers'

incentives to cyberloaf. As will be described soon, for the sake of simplicity, the present paper considers a policy to reduce material

returns from shirking deterministically in the workplace in the experiment.

Knez and Simester (2001) argued that work groups may voluntarily strengthen mutual monitoring within their groups to obtain a bonus

through achieving a firm‐level goal.

Moonlighting is increasingly common in some countries because it is encouraged by the government. For example, lifetime employment

was a common practice in Japan traditionally. However, the Japanese Ministry of Health, Labour and Welfare published the “Guidelines

for Promotion of Side Work”and deleted the description of prohibition of subsidiary business from “The Model Rules of Employment”

in 2018.

This very rarely happened in the experiment. All three members submitted the same answers in 96.9% of teams' submissions in the

experiment (3176 out of 3278 completed tasks in the 62 experiment sessions). The authors read through all the communication dialogues

and their submitted answers, and found that almost all disagreements are errors or typos. The mean number and the mode of

disagreements across all teams that disagreed were, respectively, 1.72 and 1. The size of the error rate is unsurprising because the average

number of attempts for these teams was 24.14 questions, above the average of 17.81 for the experiment, which might increase potential

errors in typing.

The authors read through the communication dialogues and found no team to have broken the anonymity rule.

The practice phase and the main task‐solving phase are called “phase 1”and “phase 2”in the experiment instructions.

To avoid cognitive overload, subjects are provided with instructions for the practice phase only at the beginning of the experiment.

Instructions for the main task‐solving phase are distributed once the practice ends. Such gradual learning approach is often taken in

experiments (e.g., Ertan et al., 2009; Kamei & Tabero, 2022; Kamei et al., 2015).

KAMEI and TABERO

This return can be thought of as material returns that can be obtained from shirking in the real workplace. Shirkers may build their

social network using social media or by exchanging e‐mails during work time, develop skills to benefit future job prospects, complete

personal tasks, or even moonlight privately as in the real‐world example described in the introduction of the paper. Such activities may

not only provide intrinsic satisfaction but may also provide material benefits. A similar designing approach was chosen by Kamei and

Markussen (2023) where an activity alternative to solving a real‐effort task is to watch a funny video. Subjects in Kamei and Markussen

(2023) received a small return per minute watching the videos.

As discussed later, the present paper uses the “weights‐based identification strategy”proposed by Dal Bó et al. (2019) to identify the

dividend of democracy. The requirement to use this method is that teams' types are independently drawn and their behavior only

depends on the team's own type. It is therefore essential to avoid dynamic interactions across teams including information feedback.

While another realistic voting method is a unanimity rule (consensus), this study adopted majority voting because the interpretation of

data becomes complex when the unanimity rule is in use as it possibly involves strategic voting among voters (e.g., Battaglini et al., 2010;

Kamei, 2019a).

The average number of correct answers and average per member working/shirking time by institutional condition can be found in

Table 3.

Each estimate was calculated using 62 sessions randomly drawn from the set of the original 62 sessions.

An analysis in Section 3.2 suggests that workers in the ENDO treatment did not accumulate fatigue with a higher work pace, as they

instead increased the time spent in the Game screen.

A team cannot complete a task while some member is shirking. Such shirking is also interpreted as maliciousness or lack of team spirit

towards members who are motivated and are waiting for the shirker's input to find the answer.

The high difficulty in finding answers to the real‐effort task is a crucial feature of the experiment, which was intentionally designed.

Notice that if the tasks were easy, worker subjects would work hard with small output elasticity of incentive changes in this kind of real‐

effort experiment (Corgnet et al., 2015; Erkal et al., 2018). A challenging real‐effort task and an availability of alternative activities

(Tetris) were thus carefully incorporated in the design to make the output elasticity of incentives sufficiently large.

Material incentives did matter for workers' effort choices. In the EXO treatment, the four teams for which task‐solving was privately

optimal worked on counting on average 31.80 mins, which is significantly larger at two‐sided p= .0015 than the average work time by

the other 87 teams where gaming was privately optimal (which was 23.12 mins)—see Supporting Information Table C1.

The effect of the reduction policy was apparently strong in the Endo treatment (see Table 3for the numbers). The average number of

correct answers in the Endo treatment was significantly larger with than without the reduction policy at two‐sided p= .001*** (.0020***)

according to the bootstrap method (a Mann–Whitney test). However, this strong effect is just due to selection. The difference was not

significant at two‐sided p= .388 when using the bootstrap method with the distribution of votes in the population being the weights

following Dal Bó et al. (2019). Recall that democracy enhanced work productivity in the experiment similarly regardless of whether the

policy was imposed or not (Result 2), whose aspect makes the effect of the policy in itself small.

For example, see the following BBC news: https://www.bbc.co.uk/news/business-64669987 (last accessed on April 21, 2023).

REFERENCES

Abeler, J., Falk, A., Goette, L., & Huffman, D. (2011). Reference points and effort provision. American Economic Review,101(2), 470–492.

Aimone, J. A., Iannaccone, L. R., Makowsky, M. D., & Rubin, J. (2013). Endogenous group formation via unproductive costs. Review of

Economic Studies,80(4), 1215–1236.

Alchian, A., & Demsetz, H. (1972). Production, information costs, and economic organization. American Economic Review,62(5), 777–795.

Araujo, F. A., Carbone, E., Conell‐Price, L., Dunietz, M. W., Jaroszewicz, A., Landsman, R., Lamé, D., Vesterlund, L., Wang, S. W., &

Wilson, A. J. (2016). The slider task: An example of restricted inference on incentive effects. Journal of the Economic Science

Association,2,1–12.

Battaglini, M., Morton, R. B., & Palfrey, T. R. (2010). The swing voter's curse in the laboratory. Review of Economic Studies,77(1), 61–89.

Bednar, J., Chen, Y., Liu, T. X., & Page, S. (2012). Behavioral spillovers and cognitive load in multiple games: An experimental study. Games

and Economic Behavior,74,12–31.

Bergh, D. D., Connelly, B. L., Ketchen, Jr., D. J., & Shannon, L. M. (2014). Signalling theory and equilibrium in strategic management

research: An assessment and a research agenda. Journal of Management Studies,51(8), 1334–1360.

Bergstrom, T., Blume, L., & Varian, H. (1986). On the private provision of public goods. Journal of Public Economics,29,25–49.

Bisetti, E., Tengelsen, B., & Zetlin‐Jones, A. (2022). Moral hazard in remote teams. International Economic Review,63(4), 1595–1623.

Bolton, P., & Dewatripont, M. (2004). Contract theory. MIT Press.

Brekke, K. A., Hauge, K. E., Lind, J. T., & Nyborg, K. (2011). Playing with the good guys. A public good game with endogenous group

formation. Journal of Public Economics,95(9–10), 1111–1118.

Chaudhuri, A. (2011). Sustaining cooperation in laboratory public goods experiments: A selective survey of the literature. Experimental

Economics,14(1), 47–83.

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement,20,37–46.

KAMEI and TABERO

Connelly, B. L., Certo, S. T., Ireland, R. D., & Reutzel, C. R. (2011). Signaling theory: A review and assessment. Journal of Management,

37(1), 39–67.

Corgnet, B., Hernán‐González, R., & Schniter, E. (2015). Why real leisure really matters: Incentive effects on real effort in the laboratory.

Experimental Economics,18, 284–301.

Dal Bó, P. (2010). Experimental evidence on the workings of democratic institutions. In S. Galiani, & I. Sened (Eds.), Institutions, property

rights, and economic growth: The legacy of Douglass North. Cambridge University Press.

Dal Bó, P., Foster, A., & Putterman, L. (2010). Institutions and behavior: Experimental evidence on the effects of democracy. American

Economic Review,100(5), 2205–2229.

Dal Bó, P., Foster, A., & Kamei, K. (2019). The democracy effect: A weights‐based identification strategy [NEBR Working Paper 2 5724].

Deci, E., & Ryan, R. (1985). Intrinsic motivation and self‐determination in human behavior. Plenum.

Deci, E. L., & Ryan, R. M. (2000). The ‘what’and ‘why’of goal pursuits: Human needs and the self‐determination of behavior. Psychological

Inquiry,11(4), 227–268.

Doellgast, V., & Marsden, D. (2019). Institutions as constraints and resources: Explaining cross‐national divergence in performance

management. Human Resource Management Journal,29, 199–216.

Elfenbein, D. W., Fisman, R., & Mcmanus, B. (2012). Charity as a substitute for reputation: Evidence from an online marketplace. Review of

Economic Studies,79(4), 1441–1468.

Erkal, N., Gangadharan, L., & Koh, B. H. (2018). Monetary and non‐monetary incentives in real‐effort tournaments. European Economic

Review,101, 528–545.

Ertan, A., Page, T., & Putterman, L. (2009). Who to punish? Individual decisions and majority rule in mitigating the free rider problem.

European Economic Review,53, 495–511.

Falk, A., Huffman, D., & Mierendorff, K. (2006). Incentive properties and political acceptability of workfare: Evidence from real effort

experiments [Working Paper].

Fehr, E., & Schmidt, K. (2006). The economics of fairness, reciprocity and altruism—experimental evidence and new theories. In S.‐G. Kolm,

& J. M. Ythier (Eds.), Handbook of the Economics of Giving, Altruism and Reciprocity (pp. 615–691). North Holland Publishing Co.

Fischbacher, U., & Gächter, S. (2010). Social preferences, beliefs, and the dynamics of free riding in public goods experiments. American

Economic Review,100(1), 541–556.

Fischbacher, U., Gächter, S., & Fehr, E. (2001). Are people conditionally cooperative? Evidence from a public goods experiment. Economics

Letters,71(3), 397–404.

Gibbons, R., & Henderson, R. (2013). What Do Managers Do? (Chapter 17) Exploring Persistent Performance Differences among Seemingly

Similar Enterprises. In R. Gibbons, & J. Roberts (Eds.), The Handbook of Organizational Economics. Princeton University Press.

Gintis, H., Smith, E. A., & Bowles, S. (2001). Costly signaling and cooperation. Journal of Theoretical Biology,213(1), 103–119.

Grimm, V., & Mengel, F. (2009). Cooperation in viscous populations—Experimental evidence. Games and Economic Behavior,66(1),

202–220.

Hagen, E. H., & Bryant, G. A. (2003). Music and dance as a coalition signaling system. Human Nature,14(1), 21–51.

Hawkes, K., & Bliege Bird, R. (2002). Showing off, handicap signaling, and the evolution of men's work. Evolutionary Anthropology: Issues,

News, and Reviews,11(2), 58–67.

Holmstrom, B. (1982). Moral hazard in teams. The Bell Journal of Economics,13(2), 324–340.

Hume, D. (2000 [1739]). A Treatise of Human Nature. Oxford University Press.

Iannaccone, L. R. (1992). Sacrifice and stigma: Reducing free‐riding in cults, communes, and other collectives. Journal of Political Economy,

100(2), 271–291.

Ichniowski, C., Shaw, K., & Prennushi, G. (1997). The effects of human resource management practices on productivity: A study of steel

finishing lines. American Economic Review,87, 291–313.

Jensen, T., & Markussen, T. (2022). Group size, signaling and the effect of democracy. Journal of Economic Behavior & Organization,187,

258–273.

Kamei, K. (2016). Democracy and resilient pro‐social behavioral change: An experimental study. Social Choice and Welfare,47(2), 359–378.

Kamei, K. (2019a). Cooperation and endogenous repetition in an infinitely repeated social dilemma. International Journal of Game Theory,

48(3), 797–834.

Kamei, K. (2019b). The power of joint decision‐making in a finitely‐repeated dilemma. Oxford Economic Papers,71(3), 600–622.

Kamei, K., & Markussen, T. (2023). Free riding and workplace democracy—heterogeneous task preferences and sorting. Management

Science,69, 3884–3904.

Kamei, K., Putterman, L., & Tyran, J. R. (2015). State or nature? Endogenous formal versus informal sanctions in the voluntary provision of

public goods. Experimental Economics,18,38–65.

Kamei, K., & Tabero, K. (2022). The individual‐team discontinuity effect on institutional choices: Experimental evidence in voluntary public

goods provision [Keio‐IES Discussion Paper Series 2022‐015]. Institute for Economics Studies, Keio University.

Knez, M., & Simester, D. (2001). Firm‐wide incentives and mutual monitoring at continental airlines. Journal of Labor Economics,19,

743–772.

Ledyard, J. (1995). Public goods: A survey of experimental research. In J. Kagel, & A. Roth (Eds.), Handbook of experimental economics (pp.

111–194). Princeton University Press.

KAMEI and TABERO

Luhan, W. J., Kocher, M. G., & Sutter, M. (2009). Group polarization in the team dictator game reconsidered. Experimental Economics,12,

26–41.

Marschak, J., & Radner, R. (1972). Economic Theory of Teams. Yale University Press.

McNamara, T., Meloso, D., & Puncheva‐Michelotti, P. (2022). ‘You are free to choose …are you?’Organisational punishment as a

productivity incentive in the social science literature. Human Relations,75, 322–348.

Melizzo, P., Carpenter, J., & Matthews, P. H. (2014). Workplace democracy in the lab. Industrial Relations. Journal,45(4), 313–328.

Niederle, M., & Vesterlund, L. (2007). Do women shy away from competition? Do men compete too much? The Quarterly Journal of

Economics,122(3), 1067–1101.

Prendergast, C. (1999). The provision of incentives in firms. Journal of Economic Literature,37(1), 7–63.

Rousseau, J.‐J. (2009 [1755]). Discourse on the Origin of Inequality. In P. Coleman (Ed.), F. Philip (Trans.). Oxford University Press.

Samuelson, P. A. (1954). The pure theory of public expenditure. The Review of Economics and Statistics,36(4), 387–389.

Smith, E. A., & Bliege Bird, R. (2000). Turtle hunting and tombstone opening. Evolution and Human Behavior,21(4), 245–261.

Smith, E., & Bliege Bird, R. (2005). Costly signaling and prosocial behavior. In H. Gintis, S. Bowles, R. Boyd, & E. Fehr (Eds.), Moral

sentiments and material interests: The foundations of cooperation in economic life (pp. 115–148). MIT Press.

Sobel, J. (2005). Interdependent preferences and reciprocity. Journal of Economic Literature, XLIII, 392–436.

Sosis, R., & Alcorta, C. (2003). Signaling, solidarity, and the sacred: The evolution of religious behavior. Evolutionary Anthropology: Issues,

News, and Reviews,12(6), 264–274.

Sosis, R., & Bressler, E. R. (2003). Cooperation and commune longevity: A test of the costly signaling theory of religion. Cross‐Cultural

Research,37(2), 211–239.

Spence, M. (1973). Job market signaling. The Quarterly Journal of Economics,87(3), 355–374.

Sutter, M., Haigner, S., & Kocher, M. G. (2010). Choosing the carrot or the stick? Endogenous institutional choice in social dilemma

situations. Review of Economic Studies,77, 1540–1566.

Thöni, C., & Volk, S. (2018). Conditional cooperation: Review and refinement. Economics Letters,171,37–40.

Tyran, J. R., & Feld, L. P. (2006). Achieving compliance when legal sanctions are non‐deterrent. The Scandinavian Journal of Economics,

108(1), 135–156.

Zelmer, J. (2003). Linear public goods experiments: A meta‐analysis. Experimental Economics,6, 299–310.

SUPPORTING INFORMATION

Additional supporting information can be found online in the Supporting Information section at the end of this article.

How to cite this article: Kamei, K., & Tabero, K. (2023). Free riding, democracy, and sacrifice in the workplace:

Evidence from a real‐effort experiment. Journal of Economics & Management Strategy,1–21.

https://doi.org/10.1111/jems.12570

KAMEI and TABERO

Employee ownership and workplace democracy: Antidotes to labour market monopsony?

Article

Mar 2025

Purpose We show that employee ownership is more efficient than control by external capital owners/employers. This complements the empirical evidence for benefits of employee ownership surveyed by Mygind and Poulsen (2021), Kruse (2022) and Dow (2003), and the normative political case for democratising work made by Ellerman (1975, 2022), Ferreras et al. (2022), Piketty (2022) and others. Of course, efficiency issues are usually important in economic evaluation. Design/methodology/approach Worker mobility or “exit” is generally costly, so employers with residual control have monopsony power to exploit workers with non-contractible job utility – who are thus less than perfectly mobile and, in the absence of collective bargaining, lack countervailing “voice”. Findings The potential for wasteful conflict and exploitation is inherent in the employment relationship, and socially optimal effort is unlikely to be achieved. We show that economic efficiency in a “sticky” world (Banerjee and Duflo, 2019) with imperfect information and incomplete contracting actually requires residual control by workers rather than just capital-labour parity in “democratic socialism”, so labour should hire capital rather than vice versa. Originality/value The “labour hires capital” allocation of rights contrasts with the traditional power of capital-owning employers who claim the firm’s residual income and control of hired employees. Such shareholder primacy not only deprives employees of their rights of self-determination and generates conflict, but also, and less obviously, generally fails to attain the efficient effort-output trade-off.

Free Riding and Workplace Democracy: Heterogeneous Task Preferences and Sorting

Article

Full-text available

Sep 2022
MANAGE SCI

A novel laboratory experiment is used to show that mismatching between task preferences and task assignment strongly undermines worker performance and leads to free riding in teams. Unlike prior experiments using real effort tasks, task preferences are elicited from all workers. Under team-based remuneration (revenue sharing), free riding is significant, but this effect is largely driven by those working on undesired tasks. Workers’ endogenous sorting into tasks improves productivity as it mitigates task mismatching although workers’ task selection per se has only small effects on work performance and effort provisions beyond the positive sorting effects. This paper was accepted by Yan Chen, behavioral economics and decision analysis.

‘You are free to choose…are you?’ Organisational punishment as a productivity incentive in the social science literature.

Article

Full-text available

Mar 2021

Authors' note: The lack of recognition of the effectiveness of punishment on employee productivity blinds the need for action to protect workers’ welfare. ################################################################ Abstract: This paper investigates the theoretical and empirical relationships between organisational punishment and productivity. We do so by highlighting the contributions of two academic fields to this topic: management and economics. We underscore the many common theoretical and empirical grounds across management and economics. We heighten, in particular, how motivation and learning theories have contributed to the development of both theoretical and empirical research on this topic. This paper also argues that this debate could be significantly advanced if insights stemming from industrial relations and labour process theory were also considered since these disciplines have traditionally focussed on macro-issues such as how changes in the economic/institutional contexts may affect the likelihood that organisations will resort to punishment. In order to foster future research on this topic, three research themes were developed: a) freedom of choice and the role of contract completeness; b) perception of punishment, monitoring, and productivity; and c) punishment, productivity and exogenous variables.

Cooperation and Endogenous Repetition in an Infinitely Repeated Social Dilemma

Article

Full-text available

Sep 2019
INT J GAME THEORY

Kenju Kamei

Exogenously imposed infinite repetition is known to mitigate people’s uncooperative behaviors in dilemma situations with partner matching through personal enforcement. One as yet unanswered question is whether people collectively choose to interact with each other under the partner matching condition when there exists an alternative possibility under random matching. In an indefinitely repeated public goods game framework, I let subjects democratically choose whether to (i) play with pre-assigned specific others for all rounds or to (ii) play with randomly matched counterparts in every round. The experimental results revealed that most groups collectively opt for the partner matching protocol. The data also indicated that groups achieve a higher level of cooperation when they democratically select the partner matching protocol by voting, relative to when the same option is exogenously imposed. These findings imply that people’s equilibrium selection may be affected by how the basic rules of games are introduced (endogenously or exogenously). The paper provides further evidence to suggest that the positive effect of democratic decision-making is stronger when the majority voting rule, rather than the unanimity rule, is applied.

Institutions as constraints and resources: Explaining cross‐national divergence in performance management

Article

Full-text available

Sep 2018
Hum Resource Manag J

This article compares performance management practices in call centres from four telecommunications firms in the United Kingdom, France, Denmark, and Germany. Findings show that different combinations of institutional constraints, such as strong job security protections, and participation resources supporting worker voice were influential in shaping choices among policies to motivate and discipline workers. Performance management most closely approached a high‐involvement model where both constraints and resources were high, where worker representatives were able both to restrict management's use of sanctions and to establish procedures that improved the perceived fairness of incentives. Findings contribute to debates concerning the role of contextual factors in the design and effectiveness of HRM.

The democracy effect: A weights-based estimation strategy

Article

Apr 2024
J ECON BEHAV ORGAN

Production, information costs, and economic organization

Chapter

Sep 2009

This book brings together classic writings on the economic nature and organization of firms, including works by Ronald Coase, Oliver Williamson, and Michael Jensen and William Meckling, as well as more recent contributions by Paul Milgrom, Bengt Holmstrom, John Roberts, Oliver Hart, Luigi Zingales, and others. Part I explores the general theme of the firm's nature and place in the market economy; Part II addresses the question of which transactions are integrated under a firm's roof and what limits the growth of firms; Part III examines employer-employee relations and the motivation of labor; and Part IV studies the firm's organization from the standpoint of financing and the relationship between owners and managers. The volume also includes a consolidated bibliography of sources cited by these authors and an introductory essay by the editors that surveys the new institutional economics of the firm and issues raised in the anthology.

MORAL HAZARD IN REMOTE TEAMS

Article

Apr 2022

We re‐examine the ability of teams to credibly self‐impose group punishments and prevent free‐riding when individual inputs are unobservable. We formulate self‐imposed group punishments as performance under‐reporting by the team. While under‐reporting is not credible in a static game, we show that simple strategies can sustain under‐reporting in a repeated game, and that the threat of under‐reporting improves welfare only if team members' preferences between shirking and team output consumption are non‐separable. Our results suggest that self‐assessments can replace increased managerial monitoring in remote work environments. This article is protected by copyright. All rights reserved

Group size, signaling and the effect of democracy

Article

Jul 2021
J ECON BEHAV ORGAN

A number of papers document that rules governing social dilemmas work better when implemented democratically than when imposed from above (the “effect of democracy”). This paper presents a theoretical model of the effect of democracy and uses laboratory experiments to test a key prediction emerging from the model, namely that the effect of democracy is stronger in small than in large communities. Results from a prisoner's dilemma experiment show that an effect of democracy is present in groups of all sizes but decreases strongly and becomes less persistent as the number of group members increases. In some respects, therefore, democracy appears to work best in small groups.

Discourse on the Origin of Inequality

Book

Apr 2009

Jean-Jacques Rousseau

In his Discourses (1755), Rousseau argues that inequalities of rank, wealth, and power are the inevitable result of the civilizing process. If inequality is intolerable - and Rousseau shows with unparalledled eloquence how it robs us not only of our material but also of our psychological independence - then how can we recover the peaceful self-sufficiency of life in the state of nature? We cannot return to a simpler time, but measuring the costs of progress may help us to imagine alternatives to the corruption and oppressive conformity of modern society. Rousseau's sweeping account of humanity's social and political development epitomizes the innovative boldness of the Englightment, and it is one of the most provocative and influential works of the eighteenth century. This new translation includes all Rousseau's own notes, and Patrick Coleman's introduction builds on recent key scholarship, considering particularly the relationship between political and aesthetic thought.

Moral Hazard in Remote Teams

Article

Jan 2020

Free riding, democracy, and sacrifice in the workplace: Evidence from a real-effort experiment

Abstract

Recommended publications

Democracy and resilient pro-social behavioral change: an experimental study

Free Riding and Workplace Democracy: Heterogeneous Task Preferences and Sorting

The perils of democracy

Cooperation and Endogenous Repetition in an Infinitely Repeated Social Dilemma