Master Thesis
Financial Incentives, Public Recognition and Negative
Externality-What’s the effect of their interaction on people’s performance?
Name: Zacharoula Papanikolaou
Student Number: 11375728
Study Track: Business Economics - Managerial Economics and Strategy
ECTS:15
Supervisor: Jeroen van de Ven
2
Statement of Originality
This document is written by Student Zacharoula Papanikolaou who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Acknowledgments
I gratefully acknowledge my supervisor, dr. Jeroen van de Ven of the Faculty of Economics and Business at the University of Amsterdam for his valuable guidance and his insightful comments that helped me complete my thesis.
I would also like to thank my parents, Vangelis and Xanthi, and my sister, Marina, for their endless love and support.
3
Abstract
This study seeks to address whether people underperform knowing that their performance harms their peers. The results, obtained from a controlled experiment, showed that, in the current setting, negative externality has a zero effect when people receive monetary incentives and a positive effect when they receive financial incentives and public recognition. Furthermore, I study the performance effect of public recognition when it is offered as an additional motivator in two payment schemes, that is financial incentives and financial incentives in the presence of negative externality. In either case, public recognition has no impact on effort provision. Finally, the comparison of the two genders indicated significance difference due to negative externality effect only when subjects are offered with monetary incentives. Additionally, substantial gender difference was found due to public recognition effect only when the latter is added to a compensation scheme that includes monetary incentives together with negative externality.
Keywords: financial incentives, public recognition, negative externality, controlled experiment
4
Table of Contents
1. Introduction 5
2. Literature review 7
3. Methodology 10
3.1 Treatments and groups 10
3.2 Task 11
3.3 Process 11
3.4 Incentives 12
3.5 Formulation of hypotheses 13
3.6 Experimental design in hindsight 14
4. Results 18
4.1 The effect of negative externality 18 4.2 The effect of public recognition 19 4.3 Gender differences - the effect of negative externality 21 4.4 Gender differences - the effect of public recognition 22
5. Discussion 23
5.1 Practical Implications 23
5.2 Limitations and further research 25
6. Conclusion 26
References 28
Appendix A:Robustness check 30
5
1. Introduction
During the last two decades, one of the topics having received special attention in organizational economics is investigating the effects of different types of incentives provided to agents (Fehr and Falk, 2002; Ashraf et al, 2014). This is of great importance to managers who seek ways in order to motivate and optimize workers’ performance using successful award systems (Danish and Usman, 2010). In this context, apart from monetary motivators, many work environments have also incorporated non-financial elements in their incentive schemes, such as public recognition for high achieving persons (Kosfeld and Neckermann, 2011; Lourenço, 2015).
Even though a substantial body of research has documented the successful results of both financial incentives and public recognition, it is not yet known that their favoring effects hold under any condition. For instance, when one’s higher performance has a negative impact on their colleagues’ payment, top performers may either feel guilty about their peers’ harassment (Hauge, 2016) or in fear of being socially stigmatized they may not want to be revealed (Fehr and Falk, 2002; Rege and Telle, 2004). In that case, providing them with financial incentives or social distinction could actually decrease performance.
Within this framework, the goal as well as the contribution of the first research question is to investigate whether people underperform when a kind of negative externality is also included in the compensation scheme that they are involved in. Two different conditions are applied in which different kind of incentives are offered. In the first condition people are given only monetary incentives, in the form of piece rates, while in the second one a bundle of piece rates and public recognition is provided. The two motivators that are selected are commonplace in recent work environments (Lourenço, 2015), therefore, their further study is a crucial topic.
The second aim of my work is to further extend current knowledge regarding the motivating power of public recognition. There are many studies that highlight the positive effects of social distinction in work environments (Bradler et al., 2016; Kosfeld and Neckermann, 2011), nevertheless, in most of these cases social acknowledgement is applied as the only incentive. Within this in mind, this paper aims to determine the performance effect of public recognition when it is provided as an additional incentive to two different payoff structures. The first payoff structure
6
includes piece rates, while the second one includes piece rates combined with a kind of negative externality. Consequently, the second research question is motivated by the concern to broaden the existing literature about the public recognition effect when it is applied as an extra incentive to certain compensation schemes. Finally, this paper provides additional evidence about how males and females behave with respect to the two research questions mentioned above.
The results of the present study are established through a controlled experiment and indicate that subjects, in total, respond the same when they are provided with piece rates and their performance has a negative impact on their peers. On the other hand, they increase their performance when they are given both piece rates and public recognition while at the same time their performance harms the others. Regarding the second research question, the results revealed that being singled out for high achievements does not affect performance. This finding holds either when people receive only piece rates or when piece rates are combined with a kind of negative externality. Furthermore, a comparison of differences reveals that negative externality affects only men’s performance when they are paid with piece rates, whereas women’s performance remains unaffected. On the opposite, neither men nor women perform differently due to the negative externality effect when they are given piece rates and public recognition. Additionally, the analysis did not show any significant gender difference due to public recognition effect when they receive piece rates however, significant difference is found when people are given piece rates in the presence of negative externality. In that case, women respond positively to public recognition, whereas men are not affected.
This paper is organized as follows. Section 2 gives a brief overview of the literature, related to my research questions. In section 3 the experimental setting is presented in detail while hypotheses are also formulated. In the same part, the experimental design is further analyzed, indicating how, in this setting, negative externality can lead to a different equilibrium to what is expected. Section 4 reports the results emerged from the data and section 5 presents some practical implications drawn from the results as well as possible limitations of the current study and suggestions for future research. Main conclusions, derived from this study are summarized in section 6.
7
2. Literature review
Large literature exists addressing the positive effect of monetary incentives on agents’ performance. Shearer (2004) for instance, by implementing a field experiment, established that monetary incentives helped improve planters’ productivity by 20% when they were paid with piece rates compared to fixed wages. In a similar way, Banker et al. (2000) discuss the concept of performance-based compensation scheme by analyzing the sales of 3,776 employees from ten retail outlets. Their findings suggest that due to the new payment plan, organization’s productivity went up not only because employees were induced to work harder but also because the company appealed to more productive agents.
With respect to social acknowledgement, related work by Kosfeld and Neckermann (2011) provides empirical evidence regarding the positive effect of social distinction on employees’ performance. The authors conducted a field experiment where participants were split in groups of 9.4 people on average and asked to execute a task. In the award treatment, participants knew in advance that the managing director of the organization they worked for was going to personally offer a congratulatory card to the two people, in each group, having exerted more effort. The results in award treatment indicated an increase in people’s performance by 12% compared to control treatment. The results were also similar even in the case of the unannounced public recognition according to Bradler et al. (2016). In this study more than 300 people were separated in groups of eight and worked individually receiving a flat wage. Based on their first two-working-hour performance participants were provided with unexpected, thank-you cards signed by the head of the research institute that the experiment had been organized. In all treatments, participants increased their performance however, the effect was even stronger when only the best three employees were recognized than when either all group members or the single-best employee got awarded. Interestingly, not recipients were those who performed remarkably higher compared to subjects who received the cards.
Even though the existing literature shed light on the positive effects of monetary incentives and public recognition, in none of the reviewed papers, one’s payment depends on others’ effort, thus, they do not address how employees perform when their performance has a negative impact on their peers. One exception is Bandiera et al. (2005) who examined whether workers internalize the negative externality they
8
cause to their peers by recording fruit pickers’ productivity, given relative incentives and piece rates. The results showed that in the first scheme, where workers' daily pay was based on their individual productivity divided by the average productivity, workers’ performance was lower compared to the second scheme, where payment depended exclusively on individual productivity. This paper provides well-grounded evidence towards understanding the performance effect of negative externality however, it differs with my experimental design since it does not examine the impact of negative externality when it is incorporated into piece rates which is one of the goals of this study.
Another paper that approaches the topic of top performers empirically is the work by Koch and Metcalfe (2011). In this paper, the authors collected self-reported data by asking participants to record every time they felt a part of an upward comparison and other relevant information such as their social bond with the other party. Through an empirical analysis, they reached the conclusion that participants’ concerns about the outperformed persons’ responses are associated with lower positive affect and higher negative affect. They also found that as high-achieving persons do care about outperformed persons’ reaction, more negative feelings such as low self-esteem and unpleasant mood are expected for the outperforming party. However, as any other study based on survey data, it may lack validity (Spector, 1994) since it only records subjects’ personal experiences without observing their behavior. Furthermore, it only describes people’s concerns about the outperformed parties but it does not investigate whether these feelings could make them underperform.
In the same framework regarding the topic of best performers, Exline et al. (2004) state that “when outperformers pose a threat to other people, their social bonds may suffer, with results ranging from awkward social encounters to rejection or retaliation by envious persons.” Similarly, Kim and Glomb (2014) through a two-study paper claim that top performers are more likely to be victimized in their workplace as a result of their coworkers’ envy. Additionally, Exline and Lobel (1999) present a framework labeled “sensitivity about being the target of a threatening upward comparison (STTUC)”, under which high performers experience distress when they believe that outperformed people are making a social comparison. They argue that strategies such as modest self-presentation and appeasement can be used to moderate STTUC, however, no empirical evidence is provided to estimate whether top performers adopt these methods in practice so as to prevent peers’ dissatisfaction.
9
As a result, important questions such as whether employees would be willing to sacrifice their own payment so as to help their peers earn more money or whether they would underperform and subsequently lose the title of best performer in order to avoid being the “target” remain unexplored. Consequently, my contribution to existing literature is to empirically investigate this research question by conducting a controlled experiment.
With respect to the second research question, that is if social distinction affects performance when people are already given piece rates and piece rates in the presence of negative externality, the papers by Bradler et al. (2016) and Kosfeld and Neckermann (2011) that are analyzed above, highlight the beneficial effect of being socially acknowledged. Nevertheless, both studies use public recognition as the only incentive and no interaction with any monetary reward occurs. The only paper that I have found closest to my study is the work by Lourenço (2015), which provides a new insight by investigating the interaction effect between monetary incentives and public recognition. In this study, the author, through a field experiment, measured the performance of eight groups of sales representatives where participants either in financial-incentives condition or in public-distinction condition were awarded based on a three-level goal. The author concluded that there is a crowd out effect between monetary incentives and public recognition since she found a negative interaction between the two motivators.
Consequently, the present study seeks to re-examine whether public recognition crowds out the effect of monetary incentives, in a controlled experiment, by comparing subjects’ performance when they receive only piece rates and both piece rates and public recognition. In addition, I complement the existing literature by analyzing if social distinction changes effort provision when people receive piece rates combined with a kind of negative externality. Finally, none of the present papers report any gender differences regarding the two research questions, which is also an additional contribution of my own study.
10
3. Methodology
3.1 Treatments and groups
Four treatments were applied in one session. The set of subjects consisted of 120 people. All participants were first-year undergraduate students in economics. Participants were randomly assigned to each group, labeled Group A, Group B, Group C and Group D, therefore, in each group there were 30 people. Table 1 presents the number of observations per group and gender. As we can see, the two genders were almost equally represented in the total sample, while the majority of participants were male in all of the four groups except Group B.
It was a between subject experiment since each participant went through only one treatment. Throughout the paper I use the terms “Piecerates”, “Rec” and “Next” to refer to piece rates, public recognition and negative externality respectively. In the first treatment, labeled Piecerates, participants received piece rates based on their performance. In the second treatment, labeled Piecerates*Rec, participants received piece rates similar to the first treatment but this time the first five performers were publicly recognized and asked to stand up. In the third treatment as well as in the fourth treatment, labeled Piecerates*Next and Piecerates*Rec*Next respectively, participants were randomly and anonymously paired with another participant of the same group. In these two treatments participants received piece rates but there was also a kind of negative externality in terms of as partner’s 1 score increased, it had a stronger negative effect on partner’s 2 final payment and vice versa. In addition, in treatment Piecerates*Rec*Next the first five performers were also recognized and asked to stand up as in treatment Piecerates*Rec.
The reason why I decided to award the first five out of thirty people in treatments Piecerates*Rec and Piecerates*Rec*Next is to create a strong motivator that could induce participants to exert more effort having almost 17% probability to be socially recognized as top scorers. Furthermore, the provision of social distinction in those treatments did not involve any additional material benefit but only highlighting top performers’ status.
11
Table1 Number of observations per group and gender
Sample
Men
Women
Group A
30 17 13Group B
30 11 19Group C
30 16 14Group D
30 18 12Total
120 62 583.2 Task
All participants had to work on the same task. The task consisted of 35 questions taken from an IQ test. This information was also available in the instructions. The type of task was intentionally chosen, so that people would be interested to perform high and be awarded. The whole IQ test included 90 questions with an increased difficulty level from question 1 to question 90, all required mental effort and based on
different types of questions such as mathematical, logical and questions with shapes. The 35 questions selected for the task were of medium level so that participants had to exert a certain effort level in order to achieve a good score while there was the space for one to score higher by working harder. People were given ten minutes to answer the questions and got informed about time in the last five and two minutes left as well as at the end of the process. During this time, participants were monitored in order cheating to be prevented.
3.3 Process
Experiment took place in the Aristotle University of Thessaloniki so, it was conducted in Greek. Participants were randomly assigned to each group by picking a card with a desk number upon their arrival and going to the corresponding seat. Due to time and space constraints all groups worked at the same time, in the same room. The experiment started with sequentially reading the instructions for all groups starting from Group A to Group D.
In all groups participants knew from the start that the experiment consisted of two parts, a task and a questionnaire. In the first part of the instructions the rules of the task as well as the incentives given in each treatment were explicitly explained.
12
Participants had three minutes to read the instructions on their own and ask questions. Right after and for the next ten minutes they performed the task. During this time, they were not allowed to use calculators but could make their own notes.
In the second part of the experiment and before payment and announcement of best performers was implemented, each one was given a questionnaire to fill in, asking about their desk number, their gender, as well as to rate the task difficulty (from 1=easy to 5=hard) and whether they liked the task or not
(
from 1=not at all to 5=certainly yes). No questions about their age and their nationality were included since no significant diversification was expected across participants.3.4 Incentives
With regard to incentives it was clearly specified in the instructions that only two out of the total number of participants in each treatment would be selected to receive the payment. All participants in all treatments were given 10-euro initial endowment and each correct answer of the task was worth €1. Unanswered questions did not count and participants were not penalized for wrong answers.
This payment method implies that in treatments Piecerates and Piecerates*Rec, participants received the 10-euro endowment plus their compensation from solving the task, based entirely on their own score. Additionally, in treatment Piecerates*Rec there was also a type of non-financial incentive, that is the public recognition for the first five performers among their fellow students. In treatments Piecerates*Next and Piecerates*Rec*Next participants received €1 for each correct answer as in previous treatments but this time they lost €0.50 for each correct answer of their partner. Therefore, the final payment was calculated based on the initial endowment plus their own score minus their partner’s score multiplied by 0.50. This means that in case of negative payment, losses were covered by the 10-euro initial endowment. Similar to treatment Piecerates*Rec, in treatment Piecerates*Rec*Next participants were additionally motivated by publicly announcing the top five performers. In case of a tie the best performers would be randomly selected among participants with the same performance.
13
As concerns the payment process, it was implemented privately and anonymously since one envelope was distributed to all subjects. The envelope contained either the payment, for the two participants randomly selected, or it was empty for the rest.
3.5 Formulation of hypotheses
People’s decisions are driven not only by self-interest motives but also by their social preferences (Fehr and Fischbacher, 2002). This means that incentives can sometimes be counterproductive when they signal a selfish behavior that undermines people’s moral standards (Bowles, 2008). Regarding the effect of social norms on employees’ effort, Fehr and Falk (2002) make a distinction between two types of norms, that is effort-enhancing norms and effort-decreasing norms that arise according to the sign of the externality that workers create for their colleagues. This implies that employees increase their effort when this leads to positive externality for their peers and decrease their effort when this results in negative externality regarding their coworkers. In view of this, I predict that negative externality leads to a decrease in subjects’ performance:
H1: Given piece rates, participants decrease their performance in case of negative externality.
H2: Given both piece rates and public recognition, participants decrease their performance in case of negative externality.
With respect to the effect of public recognition, Fehr and Falk (2002) argue that nonmonetary motives interact significantly with financial incentives. Based on their point, people’s performance is expected to change when public recognition is given as an additional incentive to a compensation scheme that includes piece rates. Lourenço (2015) has already established that financial incentives effects and public recognition effects cancel out each other. Consequently, I can formulate the third hypothesis as follows:
H3: People decrease their performance when they are given public recognition and piece rates compared to when they are given piece rates without the effect of public recognition.
14
Regarding the public recognition effect when the initial payment scheme includes piece rates in the presence of negative externality, both papers by Fehr and Falk, (2002) and Rege and Telle (2004) underline that individuals’ behavior is affected by their desire for social approval. In particular, Rege and Telle (2004) provide evidence that subjects increase their contribution to public good when they have to reveal their identity and the amount of their contributions. In this context, Ariely et al. (2009) found that participants in their experiment want to be perceived as nice persons who do good since their decisions depended on visibility. As a result, I suppose that visibility, in the form of public recognition will lead subjects to decrease their performance since the higher they score, the less their partners earn, and subsequently the more social disapproval they receive.
H4: People decrease their performance when they are given public recognition combined with piece rates and negative externality compared to when they are given piece rates with negative externality without the effect of public recognition.
3.6 Experimental design in hindsight
In their seminal paper, Huck et al. (2012) show that social norms interact with economic incentives and bring about multiple equilibria that should be taken into account in order many empirical findings to be explained. Similarly, Fehr and Falk (2002) point out that the existence of multiple levels of equilibria has important implications since managers can affect employees’ beliefs by providing them with information about the social values and the conformity with these values. Consequently, taking into account the current setting, the goal of this part is to analyze how negative externality, contrary to my expectations, can lead to a different equilibrium in which subjects increase their effort. I also, make some suggestions how participants’ choices can be explained by their social preferences.
As mentioned in hypotheses H1 and H2, people are expected to decrease their effort when negative externality is present, given that their performance harms their peers. A possible reason is that subjects act altruistically. This idea is supported by Rotemberg (1994) who argues that altruism can lead an employee to exert more or less effort taking into consideration what benefits other workers. In this context, Andreoni and Miller (2002) found that only a quarter of those who took part in their
15
experiment were money-maximizers, whereas the rest presented a kind of altruistic behavior.
Another reason that could result in subjects’ underperformance is avoiding feeling guilt. Baumeister et al. (1994) state that guilt is one of the factors that affect human decisions and induce people to help their partners by minimizing inequalities and re-allocating emotional discomfort. This idea seems to be realistic if we examine the experimental findings by Hauge (2016). In this study, the author makes a distinction between two different sources of guilt, guilt-from-disappointment (when people do not meet others’ expectations) and guilt-from-disapproval (when people do not meet others’ moral standards) and provides evidence that subjects’ generosity in a dictator game can be explained by these types of guilt-aversion.
Turning now to the current experimental setting, there are two conditions that must be satisfied in order both subjects that are partners in treatments with negative externality to underperform. The first condition refers to the kind of social preferences that both partners must have. In particular, if both partners share the same or similar kinds of social preferences such as altruistic behavior or guilt aversion, then both are motivated to exert less effort in order to prevent peers’ dissatisfaction. On the contrary, it is obvious that if both partners prefer to provide more effort, because for example, they are driven by selfish motives such as money-maximization or efficiency concerns in the form of maximizing the payoff sum (Engelmann and Strobel, 2004), then both of them will choose to perform higher. In that case, both partners will increase their effort in equilibrium.
The second condition refers to people’s beliefs about their partner’s social preferences, given that they prefer to exert less effort themselves. That is to say, if both subjects believe that their partner gives preference to lower effort level, then both will choose to underperform, as long as they are willing to underperform themselves. On the other hand, even if both subjects prefer to work less, they are likely to increase their effort if they do not believe that their partner wants to underperform as well so as to prevent the loss due to their partner’s high score. In order to make it clearer, I use a very simple model that illustrates the utility function for two partners, 1 and 2, in both cases, that is when they act independently and when, in case of negative externality, they are anonymously paired.
16
Case 1: without the negative externality effect final payment is calculated as follows: €1 for each correct answer, plus the 10-euro initial endowment. As a result, the utility function for partner 1 is:
u
1=ln(x
1+
10
) ‐ c*x
1The optimal effort level x1* is given by differentiating the utility function with
respect to x1 and setting equal to zero. Therefore,
= 0 => x
1* =
‐
10
(1)
Similarly, the utility function for partner 2 when negative externality is absent is:
u
2=ln(x
2+
10
) ‐ c*x
2and the optimal effort level is:
x
2* = ‐
10
(2)
Case 2: when negative externality is present, final payment occurs as follows: €1 for each correct answer minus €0.50 for each of the partner’s correct answer, plus the 10-euro initial endowment. Thus, the utility function for partner 1 is:
u
1=ln(x
1‐
0.5
*x
2+
10
) ‐ c*x
1Again, the optimal effort level x1** is given by differentiating the utility
function with respect to x1 and setting equal to zero. Therefore,
= 0 => x
1**=
0.5
*x
2+
–
10
(3)
Similarly, the utility function for partner 2 when negative externality is present is:
17
and the optimal effort level is:
x
2**=
0.5
*x
1+
–
10
(4)
We observe, that in equations (1) and (2), where negative externality is absent, the optimal effort is not determined by the partner’s effort level which is the case in equations (3) and (4), where negative externality is present. In particular, in (3) and (4), as partner’s 1 effort level increases, partner’s 2 optimal effort increases too and vice versa. Thus, since the two partners choose their effort level simultaneously, both are likely to exert as much effort as possible in order to make up for the loss if they believe that their partner will choose the effort level that maximizes his utility.
This is a possible situation according to Fetchenhauer and Dunning (2009) who measured the trust level between strangers and found that participants underestimated the stranger’s trustworthiness to whom they were assigned by 30-35%. In my study, the results by Fetchenhauer and Dunning (2009) may suggest that even if people are sensitive to others’ payment, they may doubt about whether their peers do care about them as well. Assuming, for instance, that partner 1 is altruistic or guilt averse and consequently, he prefers to underperform, while partner’s 2 social preferences are unknown. If partner 1 believes that partner 2 is not driven by the same preferences, that is partner 2 is likely to increase his effort, then partner 1 may want to deviate and raise his performance in order to cover his losses due to partner’s 2 high score. Consequently, if both partners decide to provide more effort driven by this way of thinking, then in equilibrium, negative externality will lead to an increase in effort by both partners.
Future studies on the current topic are, therefore, suggested to establish whether participants do actually want to score as high as possible, despite the presence of negative externality, or whether their decision are driven by their uncertainty regarding their partner’s choice. A possible variation in the current design could be to provide participants with further information about their partner’s decision. For instance, partner 1 could be asked to choose his effort level, given that his payment is not affected by partner’s 2 performance, but, on the contrary, he can negatively affect partner’s 2 payment. As a result, partner’s 1 utility function is
u
1=ln(x
1+
10
)–c*x
1 in both case 1 and case 2, while partner’s 2 utility function is18
Therefore, the optimal effort level for partner 1 is given by equation (1) in either case, whereas the optimal effort level for partner 2 is given by equation (2) when negative externality is absent and equation (4) when negative externality is present. In practice, the comparison of partner’s 1 effort level between the two cases will show whether subjects care only about their own earnings or they choose to outperform in order to cover the loss resulting from their partner’s performance. In other words, if partner’s 1 effort remains constant in both case 1 and case 2, we can clearly conclude that subjects’ performance is not affected by negative externality.
4. Results
To evaluate the outcome of the experiment, I calculated participants’ performance in terms of the total number of their correct answers. Table2 presents the descriptive statistics of subjects’ performance per treatment and gender.
Table2 Descriptive statistics: performance per treatment and gender
Sample
Men
Women
Mean SD Mean
SD
Mean
SD
Piecerates
15.03 3.60 15.71 3.51 14.15 3.65Piecerates*Rec
15.70 4.78 16.09 4.97 15.47 4.79Piecerates*Next
16.77 4.83 19.00 5.05 14.21 3.07Piecerates*Rec*Next
17.97 3.09 18.28 2.78 17.50 3.58 Notes: mean values and SD values are rounded to the nearest hundredth. .4.1 The effect of negative externality
The design implemented in the experiment allowed me to answer my research questions. Thus, by comparing participants’ performance in treatments Piecerates and Piecerates*Next, I can test the first hypothesis, while the comparison between treatments Piecerates*Rec and Piecerates*Rec*Next tests the second hypothesis.
In order to test my hypotheses I need to execute the correct statistical test. According to McCrum-Gardner (2008) the appropriate test to compare sample means for two independent groups for a normally distributed, interval-scale variable is the t-test. In the current experiment, I want to examine whether a difference occurs
19
between the means of two similar and independent data sets with respect to performance, which is an interval-scale variable. Additionally, by doing the Shapiro-Wilk normality test in Stata software, the hypothesis about normal distribution is not rejected (p-value= 0.354). Therefore, I can proceed to execute the t-test.
As we can see from table2, the average performance for the total sample increased by 1.74 units from treatment Piecerates to Piecerates*Next. However, the difference is not statistically significant (t(58)=1.58; p-value=0.114) which implies that participants’
performance remained unaffected by negative externality when they were given piece rates. This finding does not confirm Hypothesis H1 according to which subjects were expected to decrease effort provision.
As concerns the effect of negative externality given piece rates and social distinction, subjects exerted more effort since their score went up by 2.27 units from treatment Piecerates*Rec to Piecerates*Rec*Next. Interestingly, this difference is significantly different from zero (t(58)=2.18; p-value=0.029) and declares an increase
in their performance by almost 14.5%. Consequently, hypothesis H2 is not confirmed since, in this experimental design, negative externality did not decrease subjects’ performance. On the contrary, it resulted in an increase when participants received both piece rates and public recognition. A possible explanation for this contradicting result is that subjects increased their effort in order to make up for the loss due to their partner’s high score, as presented in part 3.6.
Figure 1 demonstrates the results due to negative externality effect, under the two payment schemes applied in the experiment.
4.2 The effect of public recognition
In order to assess the effect of public recognition under the two compensation schemes, I need to compare subjects’ performance between treatments Piecerates and Piecerates*Rec as well as between treatments Piecerates*Next and Piecerates*Rec*Next. As mentioned in the previous part of the results, the most appropriate statistical test is the t-test since I want to compare sample means for two independent groups with respect to performance, which is a normally distributed, interval-scale variable (McCrum-Gardner, 2008).
With reference to the first payment scheme, participants increased their average performance by 0.67 units from treatment Piecerates to Piecerates*Rec. However, the
20
difference is not significant (t(58)=0.61; p-value=0.542), which means that social
acknowledgement did not alter people’s performance when they were already incentivized with piece rates. This result contradicts Hypothesis H3.
The same finding holds even when subjects were paid with piece rates and there was negative externality. Similarly, people’s performance went up by 1.2 units from treatment Piecerates*Next to Piecerates*Rec*Next, yet, this difference is not significant either (t(58)=1.15; p-value=0.250). Consequently, public recognition caused
no change to subjects’ performance when they were motivated with piece rates in the presence of negative externality. This finding is not consistent with Hypothesis H4.
The results due to public recognition effect, under the two compensation schemes are summarized in Figure 2.
0 0,5 1 1,5 2 2,5
piece rates piece rates & public recognition
difference in performance
The effect of negative externality
Figure 1
The vertical axis measures the differences in performance due to negative externality, under the two payment schemes, given a) piece rates, b) piece rates and public recognition. On the top of each bar, it is mentioned whether the difference is significant or not.
no sign. difference
sign. difference
21
4.3 Gender differences - the effect of negative externality
Regarding the gender differences with respect to negative externality effect, we observe that men increased their performance by 3.29 units, given piece rates, that is from treatment Piecerates to Piecerates*Next. This difference is significantly different from zero (t(31)=2.16; p-value=0.031) and represents an increase in males’
performance by 21%. On the other hand, negative externality had almost no impact on women’s performance when they were given piece rates since there was only a slight, insignificant increase of 0.06 units (t(25)=0.05; p-value=0.960) in their performance.
As to gender differences when people received both piece rates and public recognition, we observe a similar trend since both males and females performed higher by 2.19 and 2.03 units respectively from treatment Piecerates*Rec to Piecerates*Rec*Next. Nevertheless, none of the two differences is statistically significant (t(27)=1.34; p-value=0.180 for males and t(29)=1.35; p-value=0.177 for
females).
Figure 3 demonstrates the results per gender due to negative externality effect, under the two payments schemes.
0 0,5 1 1,5
piece rates piece rates & negative externality
difference in performance
The effect of public recognition
Figure 2
The vertical axis measures the differences in performance due to the effect of public recognition, given a) piece rates, b) piece rates with negative externality. On the top of each bar it is mentioned whether the difference is significant or not.
no sign. difference
no sign. difference
22
4.4 Gender differences - the effect of public recognition
In relation to the two genders separately, the analysis did not show any substantial difference due to the public recognition effect when the initial payoff structure included piece rates. Particularly, men’s performance increased insignificantly by only 0.38 units (t(26)=0.22; p-value=0.826) between treatments Piecerates and
Piecerates*Rec, whereas women performed higher by 1.32 units. Yet, this difference is not significant either (t-stat(30)=0.88; p-value=0.379).
Remarkably, I found a significant gender difference when social distinction was offered as an extra incentive to piece rates while negative externality was present. That is to say, men decreased their performance from treatment Piecerates*Next to Piecerates*Rec*Next by 0.72 units, nevertheless, this difference is not significant (t(32)=-0.51; p-value=0.610). On the other hand, women significantly raised their
performance by having on average 3.29 more correct answers (t(24)=2.49;
p-value=0.013), which declares a 23% increase in their performance.
The results per gender due to public recognition effect, under the two compensation schemes are summarized in Figure 4.
0 0,5 1 1,5 2 2,5 3 3,5
men women men women
The effect of negative externality per
gender
piece rates piece rates & public recognition sign. difference no sign. difference no sign. difference no sign. difference Figure 3The vertical axis measures the differences in performance due to negative externality effect per gender, under the two payment schemes, given a) piece rates, b) piece rates and public recognition. On the top of each bar, it is mentioned whether the difference is significant or not.
difference in performance
23
5. Discussion
5.1 Practical Implications
Contrary to expectations, my results did not support any of the two hypotheses regarding the negative externality effect. Given only piece rates, participants, in total, did not alter their performance when negative externality was present. On the other hand, when the incentive scheme included both piece rates and public recognition, subjects responded positively to negative externality with a 14.5% increase in their performance.
These findings have interesting management implications. On the one hand, negative externality does not interact with piece rates in terms of its impact on effort provision. Consequently, in a payment scheme similar to my design, organizations are neither positively nor negatively affected regarding employees’ performance, when negative externality is incorporated into piece rates. On the other hand, in this setting, organizations can profit from negative externality when it is incorporated into a
-1 -0,5 0 0,5 1 1,5 2 2,5 3 3,5
Men Women Men Women
The effect of public recognition per
gender
piece rates
piece rates & negative externality no sign. difference no sign. difference no sign. difference sign. difference Figure 4
The vertical axis measures the differences in performance due to the effect of public recognition per gender, given a) piece rates, b) piece rates with negative externality. On the top of each bar it is mentioned whether the difference is significant or not.
difference in performance
24
payment scheme that includes piece rates and public recognition since it increases workers’ performance. Remarkably, either when negative externality has a positive or no effect on performance, organizations receive important economic benefits related to the amount of money they spend on workers’ payment. That is to say, due to negative externality, the average payment for each employee is lower compared to when negative externality is absent. Particularly, in my experiment, given only piece rates, negative externality resulted in a decrease in average payment from almost €25 to around €18.5. Similarly, the provision of both piece rates and public recognition under the negative externality effect caused a decrease in average payment from €25.7 to €19.5.
With regard to public recognition effect, my results were below expectations too. When public recognition was added as an extra motivator to piece rates, subjects did not decrease their performance according to previous finding by Lourenço (2015). Particularly, being awarded as top performer had no impact on effort provision. The zero effect was also evident even when public recognition was attached to the second payment scheme, that is piece rates with negative externality. In contradiction with previous studies (Fehr and Falk, 2002; Rege and Telle, 2004) that underline the importance of being socially disapproved, my analysis indicated that subjects were indifferent to being publicly recognized for high achievements, even though their performance harms their peers.
Although public recognition was not detrimental to subjects’ effort, it did not bring any important benefits either. This implies that managers are not recommended to incorporate this intervention when the applied compensation scheme includes either piece rates or piece rates with a kind of negative externality. Furthermore, taking into account additional costs, for instance, the time of managers to award best performers and the cost of reporting (Lourenço, 2015), public recognition can bring about indirect costs to organizations. However, this conclusion does not hold for workplaces where women are overrepresented and the payoff structure consists of piece rates in the presence of negative externality, since in that case, being socially acknowledged resulted in a significant performance increase for females.
25
5.2 Limitations and further research
The data of the preset study was gathered through a controlled experiment. As a result, the compensation schemes applied in my setting can be considered exogenous since they are not associated with any special, unobserved characteristics of each firm (Paarsch and Shearer, 2000; Shearer, 2004). This implies that my results are not affected by endogeneity problems (Shearer, 2004). In addition, participants were randomly allocated to each group and all of them were on the same education level, thus, increasing the internal validity of the experiment.
Nevertheless, similar to any other controlled experiment, there are some limitations that should be mentioned. First of all, as stated in part 3.6, I am aware that in my experimental setting there are two effects, a negative one caused by negative externality towards others, and a positive one, resulting from one’s desire to cover the loss, given that their partner can harm them. A possible solution how to address the positive effect of negative externality is presented in part 3.6. Secondly, regarding the process of the experiment itself, due to time and space constraints all groups executed the task at the same time and the same place. Consequently, there is some likelihood that internal validity may have been reduced since subjects were able to understand the goal of the experiment. Regarding the generalizability of the results, Levitt and List (2007) call into question the extent to which laboratory findings provide evidence for the real world. In view of this, it is noteworthy that the findings were obtained by measuring students’ and not real employees’ performance. In addition, it was a one-shot experiment and no repeated trials occurred. This implies that the results may not be applicable to a real workplace where agents are susceptible to more intense and continuous pressure from their coworkers. Furthermore, my tight payment scheme may have imposed additional constraints on subjects’ behavior. This does not reduce the internal validity of the experiment since this element remained the same across the different manipulations but it might affect the external validity in different settings. For instance, participants may have behaved in a different way if the impact on their peers was more than €0.50 per correct answer or if they were able to pursue more money for each correct answer.
Despite those limitations, my study offers the base for additional significant research. First of all, it would be fruitful to replicate the setting but this time having each group work separately. This would further increase the internal validity and
26
trustworthiness of the results. Also, based on Ariely et al. (2009) who argue that subjects’ decisions are determined by visibility, forming pairs and announcing the two partners in public is recommended in order to find out if people choose differently when image concerns are more intense since they will have the opportunity to recognize their partners. Additionally, Bandiera et al. (2005) found that relationships among workers affect the extent that employees internalize negative externality. This statement implies that measuring the social relations between the partners and investigating how this affects their performance could be an interesting follow-up study as well. Moreover, future work should concentrate on implementing the same design varied with different amounts of money per correct answer that people receive and lose due to negative externality. Fehr and Falk (2002), for instance, suggest that, apart from the sign, the size of externality affects effort provision as well. Finally, future studies should target at investigating if the results regarding the effect of negative externality and public recognition hold in long term. According to Kohn (1993), one of the reasons that incentives and awards fail is that they create only temporary compliance, while Benabou and Tirole (2003) point out that rewards are only short term, positive motivators. That is to say, being constantly engaged in a payoff structure that includes a kind of negative externality may eventually harm performance if it is too hard for employees to address continuous disapproval and awkward reactions by their peers.
6.
Conclusion
This study identifies a setting in which negative externality has either zero or positive effect on people’s performance depending on the payment scheme which is applied. Furthermore, the experimental design enabled me to highlight the zero effect of public recognition when it is as added as an additional incentive to two different payoff structures that were examined in the experiment. Contrary to women who do not behave differently, men respond positively to negative externality given only piece rates. On the other hand, no significant difference occurs when participants are provided with both piece rates and public recognition, since none of the two genders change their performance due to negative externality. Similarly, no difference between men and women is found when apart from piece rates, top performers are also publicly recognized. Yet, the analysis revealed substantial gender difference
27
regarding the public recognition effect, when people are given piece rates combined with negative externality. In that case, women are induced to perform higher knowing that they will be publicly acknowledged for their high achievements, whereas men perform the same. Practical implications regarding the effect of negative externality and public recognition are also arise, nevertheless, given the special characteristics of my design, caution must be taken in the interpretation of the results regarding the effect of negative externality on effort provision. Despite this, the current study could be a starting point to further investigate the impact of negative externality on employees’ performance and the extent to which public recognition can be an efficient tool that motivates employees.
28
References
Andreoni, J., & Miller, J. (2002). Giving according to GARP: An experimental test of the consistency of preferences for altruism. Econometrica, 70(2), 737-753.
Ariely, D., Bracha, A., & Meier, S. (2009). Doing Good or Doing Well? Image Motivation and Monetary Incentives in Behaving Prosocially. American
Economic Review, 99 (1), 544-555.
Ashraf, N., Bandiera, O., & Jack, B. K. (2014). No margin, no mission? A field experiment on incentives for public service delivery. Journal of Public
Economics, 120, 1-17.
Bandiera, O., Barankay, I., & Rasul, I. (2005). Social preferences and the response to incentives: Evidence from personnel data. The Quarterly Journal
of Economics, 120(3), 917-962.
Banker, R. D., Lee, S., Potter, G., & Srinivasan, D. (2000). An empirical analysis of continuing improvements following the implementation of a performance-based compensation plan. Journal of Accounting and
Economics, 30(3), 315-350.
Baumeister, R. F., Stillwell, A. M., & Heatherton, T. F. (1994). Guilt: An interpersonal approach. Psychological Bulletin, 115(2), 243.
Benabou, R., & Tirole, J. (2003). Intrinsic and extrinsic motivation. The
Review of Economic Studies, 70(3), 489-520.
Bowles, S. (2008). Policies designed for self-interested citizens may undermine "the moral sentiments": Evidence from economic experiments. Science (New York, N.Y.), 320(5883), 1605-1609. doi:10.1126/science.1152110 [doi]
Bradler, C., Dur, R., Neckermann, S., & Non, A. (2016). Employee recognition and performance: A field experiment. Management Science, 62(11), 3085-3099.
Danish, R. Q., & Usman, A. (2010). Impact of reward and recognition on job satisfaction and motivation: An empirical study from Pakistan. International
29
Engelmann, D., & Strobel, M. (2004). Inequality aversion, efficiency, and maximin preferences in simple distribution experiments. The American
Economic Review, 94(4), 857-869.
Exline, J. J., & Lobel, M. (1999). The perils of outperformance: Sensitivity about being the target of a threatening upward comparison. Psychological
Bulletin, 125(3), 307.
Exline, J. J., Single, P. B., Lobel, M., & Geyer, A. L. (2004). Glowing praise and the envious gaze: Social dilemmas surrounding the public recognition of achievement. Basic and Applied Social Psychology, 26(2-3), 119-130.
Fehr, E., & Falk, A. (2002). Psychological foundations of incentives. European Economic Review, 46(4), 687-724.
Fehr, E., & Fischbacher, U. (2002). Why social preferences matter–the impact of non‐selfish motives on competition, cooperation and incentives. The
Economic Journal, 112(478), C1-C33.
Fetchenhauer, D., & Dunning, D. (2009). Do people trust too much or too little? Journal of Economic Psychology, 30(3), 263-276.
Hauge, K. E. (2016). Generosity and guilt: The role of beliefs and moral standards of others. Journal of Economic Psychology, 54, 35-43.
Huck, S., Kübler, D., & Weibull, J. (2012). Social norms and economic incentives in firms. Journal of Economic Behavior & Organization, 83(2), 173-185.
Kim, E., & Glomb, T. M. (2014). Victimization of high performers: The roles of envy and work group identification. Journal of Applied Psychology, 99(4), 619.
Koch, E. J., & Metcalfe, K. P. (2011). The bittersweet taste of success: Daily and recalled experiences of being an upward social comparison target. Basic
and Applied Social Psychology, 33(1), 47-58.
Kohn, A. (1993). Why incentive plans cannot work. Harvard Business
Review, 71(5).
Kosfeld, M., & Neckermann, S. (2011). Getting more work for nothing? Symbolic awards and worker performance. American Economic Journal:
30
Levitt, S. D., & List, J. A. (2007). What do laboratory experiments measuring social preferences reveal about the real world? The Journal of Economic
Perspectives, 21(2), 153-174.
Lourenço, S. M. (2015). Monetary incentives, feedback, and Recognition— Complements or substitutes? Evidence from a field experiment in a retail services company. The Accounting Review, 91(1), 279-297.
McCrum-Gardner, E. (2008). Which is the correct statistical test to use? British Journal of Oral and Maxillofacial Surgery, 46(1), 38-41.
Paarsch, H. J., & Shearer, B. (2000). Piece rates, fixed wages, and incentive effects: Statistical evidence from payroll records. International Economic
Review, 41(1), 59-92.
Rege, M., & Telle, K. (2004). The impact of social approval and framing on cooperation in public good situations. Journal of Public Economics, 88(7), 1625-1644.
Rotemberg, J. J. (1994). Human relations in the workplace. Journal of
Political Economy, 102(4), 684-717.
Shearer, B. (2004). Piece rates, fixed wages and incentives: Evidence from a field experiment. The Review of Economic Studies, 71(2), 513-534.
Spector, P. E. (1994). Using self‐report questionnaires in OB research: A comment on the use of a controversial method. Journal of Organizational
Behavior, 15(5), 385-392.
Appendix A: Robustness check
In this part I run two linear regressions in Stata software in order to further check the robustness of my results. In both regressions productivity is the dependent variable and is defined as the number of the correct answers out of the maximum number of the correct answers. In the first specification, the treatments of the experiment are the only regressors, while in the second specification some background characteristics, gathered from the questionnaires distributed to participants are also included. Table3 presents the outcome of the linear regressions. In both regressions, treatment Piecerates is used as the baseline.
In the first column, the coefficient of treatment Piecerates*Next is positive, yet insignificant (p-value=0.117). As a result, productivity between treatments Piecerates
31
and Piecerates*Next is not different. This finding confirms my previous results that subjects did change their performance due to negative externality when they payment scheme included piece rates.
On the other hand, coefficient of treatment Piecerates*Rec*Next is positive and statistically significant (p-value=0.001). This implies that productivity in treatment Piecerates*Rec*Next is 0.514=(0.430+0.084). Additionally, the coefficient of treatment Piecerates*Rec is positive but insignificant (p-value=0.543). This means that productivity in treatment Piecerates* Rec is 0.430, that is equal to productivity in treatment Piecerates which is given by the constant. Consequently, productivity in treatment Piecerates*Rec*Next is higher compared to treatment Piecerates*Rec. This finding is consistent with previous analysis which indicated that participants increased their performance due to negative externality when they were given a bundle of piece rates and public recognition. Furthermore, the F-test finds that the effects between the two treatments are statistically different from one another (p-value= 0.031).
With respect to public recognition effect, as reported above, the coefficient of treatment Piecerates*Rec is insignificant, therefore, there is no difference in productivity between the treatments Piecerates and Piecerates*Rec. This is in line with my previous finding that subjects did not respond differently when, apart from piece rates, they were also publicly recognized as top scorers compared to when they received only piece rates.
Finally, the F-test finds no significant difference between the effects of treatments Piecerates*Next and Piecerates*Rec*Next (p-value= 0.254). This also matches with my previous results that there was no difference in people’s performance due to public recognition when the initial compensation scheme included piece rates with negative externality.
In column 2, the gender, three dummy variables for task difficulty1 and three dummy variables for task satisfaction2 have been added as control variables. Specification 2 causes a change in the significance level of treatment Piecerates*Next (p-value=0.082) which implies that, controlling for gender and task, subjects
1
The three dummy variables are: “easy_level” for those who rated task difficulty with 1 or 2, “medium_level” for those who rated task difficulty with 3 and “hard_level” for those who rated task difficulty with 4 or 5.
2
The three dummy variables are: “negative” for those who rated task satisfaction with 1 or 2, “neutral” for those who rated task satisfaction with 3 and “positive” for those who rated task satisfaction with 4 or 5.
32
increased their performance due to negative externality, given piece rates. However, the variable is significant only at 10% significance level. Specification 2, also leads to an increase in the effect of treatment Piecerates*Rec from 1.9% to 3.2% however, it remains insignificant (p-value=0.350). R-squared increases too from 6.9% to 14.3%, while the effect of treatments Piecerates*Next and Piecerates*Rec*Next remains almost the same. Additionally, although the F-test finds no difference between treatments Piecerates*Next and Piecerates*Rec*Next (p-value=0.341) as it happens in specification 1, it also finds no difference in the effects between the treatments Piecerates*Rec and Piecerates*Rec*Next (p-value= 0.131). This result is not consistent with the finding in column 1 and implies that participants did not increase their performance due to negative externality when they received both piece rates and social acknowledgment when I control for gender and task.
Table3 The effect of treatments on productivity
(1) (2)
Piecerates reference reference Piecerates*Rec 0.019 (0.031) 0.032 (0.034) Piecerates*Next 0.050 (0.031) 0.053* (0.030) Piecerates*Rec*Next 0.084*** (0.025) 0.081*** (0.024) constant 0.430*** (0.019) 0.348*** (0.031)
controls No Gender, Task
Nr. of obs. 120 120
R2 0.069 0.143
Notes: Dependent variable: productivity. Robust standard errors are shown in parentheses. Significance levels: ***p <0.01, **p <0.05, *p <0.1
“male” was used as a control variable for gender.
“easy_level” and “medium_level” were used as control variables for task difficulty, “hard_level” was omitted to prevent mutlicollinearity.
“positive” and “neutral” were used as control variables for task satisfaction, “negative” was omitted to prevent multicollinearity.
33
Appendix B: Instructions and Questionnaire
Instructions for treatment 1
Welcome
Welcome to this experiment. Please read these instructions very carefully. You are not allowed to communicate with other participants during the experiment.
This experiment consists of 2 parts.
You can earn money in this experiment. The amount depends on your own performance. You will be given 5 pages in total that have to stay covered on your desk until the moment you will be asked to do so. Please remember to write down the number you find on your desk on the first page.
Description of part1
This part will take 10 minutes. During this time you have to solve a task that consists of 35 questions. These questions are part of an IQ test. You have to answer each question on the special space given next to each question. No extra answer sheet will be given to you. I will clearly inform you about the last 5 and the last 2 minutes as well as at the end of this part.
Each participant will receive 10-euro initial endowment. In this part each correct answer is worth €1. Unanswered questions do not count. You do not lose marks for the wrong answers. At the end of the experiment, two participants will be randomly selected for payment. Each participant is equally likely to be selected for payment. Payment will be completely anonymous. At the payment stage each participant will receive one envelope. This envelope will contain either the payoff (for those who will be selected for payment) or it will be empty (for the rest).
Brief summary for part 1
o You have to solve a task made by 35 questions in 10 minutes. o Everyone will receive 10-euro initial endowment.
o Each correct answer is worth €1.
o At the end of the experiment two participants will be randomly selected to be anonymously paid.
34
Please raise your hand if you have any questions.
Description of part2
In this part you have to fill in a questionnaire that will be given to you. As soon as you complete them, questionnaires will be collected. For the rest time and until payment is implemented you have to remain silent and not to communicate with other participants.
35
Instructions for treatment 2
Welcome
Welcome to this experiment. Please read these instructions very carefully. You are not allowed to communicate with other participants during the experiment.
This experiment consists of 2 parts.
You can earn money in this experiment. The amount depends on your own performance. You will be given 5 pages in total that have to stay covered on your desk until the moment you will be asked to do so. Please remember to write down the number you find on your desk on the first page.
Description of part1
This part will take 10 minutes. During this time you have to solve a task that consists of 35 questions. These questions are part of an IQ test. You have to answer each question on the special space given next to each question. No extra answer sheet will be given to you. I will clearly inform you about the last 5 and the last 2 minutes as well as at the end of this process.
Each participant will receive 10-euro initial endowment. In this part each correct answer is worth €1. Unanswered questions do not count. You do not lose marks for the wrong answers. At the end of the experiment, two participants will be randomly selected for payment. Each participant is equally likely to be selected for payment. Payment will be completely anonymous. At the payment stage each participant will receive one envelope. This envelope will contain either the payoff (for those who will be selected for payment) or it will be empty (for the rest).
Also, at the end of the experiment and before the payment is implemented I will publicly announce the desk numbers of the first five performers. Best performers will be asked to stand up. In case of a tie best performers will be randomly chosen among those participants with the same score.
Brief summary for part 1
36
o Everyone will receive 10-euro initial endowment. o Each correct answer is worth €1.
o At the end of the experiment two participants will be randomly selected to be anonymously paid.
o Before the payment is implemented the first five performers will be publicly announced and asked to stand up.
Please raise your hand if you have any questions.
Description of part2
In this part you have to fill in a questionnaire that will be given to you. As soon as you complete them, questionnaires will be collected. For the rest time and until payment is implemented you have to remain silent and not to communicate with other participants.
