1
Bachelor Thesis 1
st
Semester
Finance & Organization
‘’The effect of tournament incentives in the
catering industry: Experimental evidence from
the workplace’’
By:
Jurre van Berkel
Student number: 10787658
Amsterdam, 30th of January 2018
2
Statement of Originality
This document is written by Jurre van Berkel who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Table of Contents
Abstract ... 3
Introduction ... 4
Literature review ... 6
Methodology ... 12
Results... 18
Conclusion & Discussion ... 23
References ... 26
4
Abstract
This paper presents field evidence that creates new insights on the provision of tournament incentives in the catering industry – especially for waiters/waitresses in restaurants. I design and conduct a field experiment at a medium sized restaurant to study the effect of
tournament incentives on the performance of employees in the catering industry. During the tournament performance is measured by the average sales of ciabatta’s pesto per
employee. The employee who has the highest average at the end of the tournament wins a reward. The overall findings suggest that the tournament incentives increase the average performance of both extrovert and introvert employees in the catering industry.
Introduction
In almost everyone working environment there are bosses and workers. The relationship between bosses and workers, also called employers and employees, does not always go without conflicts. Conflicts arise because employers and employees have other preferences. For example, in a rational world, an employer wants to maximize profits and firm value. While an employee just wants to maximize his own utility. This may lead to some conflicts between the employers and employees. Gibbons (1998) described this conflict as the basic principal-agent model, where the employer is the principal and the employee is the agent. According to Gibbons, the principal gives the agent a reward based on his output. This reward consists, for example, of salary or a bonus and is laid down in a contract. To produce this output, the agent has to put in his own effort, which is costly for the agent. So the principal wants to maximize the firm value, while the agents want to maximize their own utility, which consist of the compensation (reward) minus the cost of effort (Gibbons, 1998, p. 2). If the utility of an agent is negative, the agent will reduce his effort to reduce the cost of effort, or otherwise he will leave the company and work elsewhere. The principal can prevent this by giving a high enough incentive to the agent, so that he will stay at the company and put in high effort to produce a high output for the principal. Gibbons (1998) assumes that the principals and the agents in his model are rational. I know that in the real world this is not always the case. Sometimes incentives do not lead to the desired effect. Gneezy, Meier, & Rey-Biel (2011) say that the effects of incentives depend on how they are designed, the form in which they are given (especially monetary or nonmonetary), how they interact with intrinsic motivations and social motivations. For every area, where explicit
5
incentives seek to change behavior, the effect can be different. Potential conflicts may arise between the direct extrinsic effect of the incentives and how these incentives can crowd out intrinsic motivations in the short run and the long run. For example, monetary incentives from principals may change how tasks are perceived by agents. If incentives are not large enough, this change in perception can lead to undesired effects on behavior. In other cases, incentives might have the desired effects in the short term, but may weaken intrinsic
motivation. Thus, once the incentives are removed, people may pursue the desired outcome less eagerly (Gneezy et al., 2011, p. 16).
In this research I want to investigate, if incentives have a positive effect on the performance of employees in the catering industry. I will test this through tournament incentives. I will design my own experiment at a real restaurant, and see if the tournament structure leads to higher performance of the employees at the restaurant. My research question will be: ‘’Do incentives in tournaments have a positive effect on the performance of
employees in the catering industry?’’
Previous research has shown that tournament incentives can lead to positive effects on the performance of employees. Previous literature shows that incentives are an
important instrument for the principal as well as for the agents. With the right incentives, principals can align the preferences of the agents with their own preferences. Bandiera et al. (2005), finds that the introduction of a tournament pay scheme can lead to an increase of productivity of the employees. Delfgaauw et al. (2013) show that a monetary reward is not always necessary to increase productivity. Non-monetary rewards can also lead to an increase in productivity. And Kapoor (2010) find that, giving the right incentives to
employees based on the preferences of the firm, has benefits for the employees as well as for the firm. But I also know, as discussed before, that incentives can have negative effects for the firm and employees (Gneezy. et al., 2011).
This research will be different from the previous ones. Because previous literature often investigate the effect of tournament incentives in the simple production industry. I will apply the tournament pay structure in the catering industry in which employees have
various tasks. All of those tasks contribute to the performance of the employees. There is still much unclear in this sector regarding to incentives. This research will give new insights in
the field of incentives in multi-tasks jobs.
6
literature more in detail. After that I will state my main hypothesis. The section methodology provides a detailed description of the data and the research design. After that a theoretical framework generating my predictions about the sales of ciabatta’s pesto under the
experimental treatment. I will also describe the empirical models I will use to examine the data and explain why I use these models in particular. In the section of the results I will discuss the empirical findings and their theoretical implications. Finally in the conclusion and discussion section, I will summarize and discuss my findings and discuss some ideas for future research.
Literature review
There is done a lot of previous research about the impact of incentives on productivity. For example by Lazear (2000). He investigates what happens to the productivity of employees when a firm switches from paying hourly wages to paying piece rates. He considers data from a natural experiment at a glass corporation, a company in the simple production industry. If the weekly pay of employees was less than their fixed wage before the switch to piece rates, they received a guaranteed wage. He finds that the average productivity
increases by 44% when the firm switches to piece rate pay. 22% due to a pure incentive effect and the other 22% due to the sorting effect, which means that new employees that are hired are more productive and the old workers with low productivity left the firm after the new reward system was imposed. Furthermore, he shows that the gains are shared with the employees, they earn a 7% higher wage after the switch to piece rates (Lazear, 2000). The results imply that productivity effects associated with the switch from hourly wages to piece rates are quite large (Lazear, 2000, p. 1359). The only downside for the company is that the cost of monitoring has increased, because under piece rate the productivity of every employee has to be monitored.
Another way to give employees an incentive to increase their productivity is by means of a tournament. I will focus more on this incentive scheme, because my own experiment will be based on a tournament incentive scheme. A tournament is a
contest/competition set up by the principal where two or more agents with equal or similar tasks and responsibilities compete for promotion, bonuses, or other rewards (Lazear & Rosen, 1981). The agent/agents with the highest output or productivity, will win this promotion or bonus. Lazear and Rosen (1981) say that employees are willing to perform
7
more when a marginal increase in effort leads to a proportional increase in utility. A principal can achieve this by adopting a tournament incentive scheme in his company. There will be a competitive setting, that could lead to an efficient compensation scheme (Lazear & Rosen, 1981). Which again could lead to higher productivity per individual. The optimal result for the principal is to design a tournament that increases the motivation of the agents to exert the optimal amount of effort that increases their output (Bull, Schotter & Weigelt, 1987). Nevertheless, tournaments could also lead to problems and disadvantages for both principal and agents. In order to win the tournament and thus win the promotion/bonus. An agent must ensure that he or she has the highest output, compared to his/her colleagues. To achieve this the agent can engage in unauthorized activities. For example, collusion,
excessive risk-taking or even sabotage and mobbing (Gürtler, 2010; Harbring, & Irlenbusch, 2011). Through these unauthorized activities, agents could increase their own output or decrease the output of colleagues. This can be a problem for the principal, because these activities often lead to a negative effect on the firm’s output (Gürtler, 2010). The principal could anticipate by punish the agents who participate in these activities. In this case, agents do not have a reason to participate in unauthorized activities, because the punishment can ensure that the agents no longer have a chance of winning the bonus. Another negative effect of the tournament model could be that an agent who did not win the tournament, will be disappointed and will be less motivated in the future interaction with the principal (Bull et al., 1987). For the principal this could lead to a decrease in output or productivity in the future. Gürtler (2010) describes the problem of collusion in tournaments. In tournaments, only the relative performance of agents is important. Therefore, employees have an incentive to collude by jointly deviating to low effort levels. In this way the output will decrease, which is not optimal for the principal. Friendships among employees can enhance the collusion problem (Gürtler, 2010). But friendship can also have positive effects in the workplace. Bandiera, Barankay, & Rasul, (2009) show that social connections in the workplace, can enhance performance of these employees.
In another research Bandiera, Barankay, & Rasul, (2005) provide evidence from a field experiment on the effects of a tournament pay structure on the productivity at a soft fruit producing company in the United Kingdom, a company in the simple production industry. As earnings depend on productivity, Bandiera et al. (2005) expect that the least productive workers would quit and more able workers would join, also called: the sorting
8
effect (Lazear 2000). They expect that this change in the composition of the workforce would strengthen the positive effect of tournaments on productivity, as the workers for whom the effect is negative would prefer to quit (Bandiera et al., 2005). The results show that the introduction of a tournament pay structure increases the average productivity by 24%.
Besides the effect of the tournament pay structure on productivity tested by
Bandiera et al. (2005), there is also another effect on productivity, of the tournament model. Delfgaauw, Dur, Sol, & Verbeke (2013), perform a field experiment in a Dutch retail chain consisting of 128 stores. About half of the stores, randomly selected, were placed in a subgroup of stores. They introduced short-term sales competitions among stores. The selected stores were divided into pools of 5 and competed for a period of 6 weeks on the basis of percentage sales growth compared to the same period the year before (Delfgaauw et al., 2013). All employees of the store with the highest sales growth over 6 weeks received a bonus of 75 euro; employees of each pool’s runner-up received 35 euro. The stores that took part in the competition received weekly feedback in the form of a poster that ranked each store in their pool on their cumulative sales growth figures (Delfgaauw et al., 2013, pp. 2-3). Parallel to the treatment described above, the other half of the stores were also placed in a subgroup of stores. The stores competed in tournaments with the same setup except for the absence of a monetary reward for winning. Delfgaauw et al. (2013) find that the
tournament with the absence of a monetary reward also have a significantly positive effect on productivity. The effect is of similar magnitude as the effect of tournaments with
monetary rewards. They suggest that there is a high symbolic value of winning a tournament, which could also explain the rise in productivity.
Quiñones, Ford, & Teachout (1995) they investigate if working experience is positive correlated with job performance. They perform a meta-analysis to examine the overall effect of work experience on performance, as well as the potential moderators of this relationship. The results of the meta-analyses revealed that the relationship between work experience and job performance was positive (Quiñones et al. 1995). This implies that employees with more working experience on average have a higher productivity then employees with less working experience. If I apply this to the tournament model, it can be concluded that employees with more working experience have a higher chance of winning the tournament (Lazear & Rosen, 1981). According to Ederer (2010) this leads to a decrease in the motivation of workers with less ability/experience. He calls this effect the motivation effect, because it
9
motivates employees who know that they have more abilities than others, to exert more effort. However, it also discourages the motivation and the effort exertion by employees with less abilities/experience (Ederer, 2010).
The previous discussed articles investigate the effect of different incentives on the productivity of employees, focusing mostly on the simple production industry. For example Lazear’s (2000) experiment considers employees in a glass corporation company, where he measures a single task performed by the workers. The same applies for the research of Bandiera et al. (2005). Their field experiment investigates employees at a soft fruit producing company. Also here, only one task is measured for the productivity of workers. But there are also a lot of multi-tasks jobs. In such jobs workers have to perform several tasks. Next I will consider the impact of incentives in multi-task jobs.
The theoretical and empirical literature on multi-task principal–agent problems is large. Most relevant for my research are the researches of Fehr and Schmidt (2004) and Kapoor (2010). Fehr and Schmidt (2004) did a laboratory experiment on a two-task principal-agent model where the performance of only one task was contractible. With this experiment they found that strong incentives can have a negative effect. If an agent is offered a piece rate for one task but no explicit incentives on the other task, then he is likely to concentrate his effort on the task that is rewarded and neglect the other task. This is also known as negative spillover effects. Fehr and Schmidt (2004) provide their own solution for this
problem, namely: ‘’It is less costly for the agent to expend effort than for the principal to pay half of the social surplus upfront to the agent. Therefore, the agent should first expend his effort and then the principal should reward him, and not the other way round’’ (Fehr and Schmidt, 2004, p. 473). According to Fehr and Schmidt (2004) agents have to move first. After they exerted their effort the principal can reward them, based on their level of effort. This prevents the negative spillover effects.
Kapoor (2010) did a research about incentive provision in multi-task jobs, especially the catering industry. This research is closely related to my research. He designed and conducted a field experiment at a large-scale restaurant in Canada. The existing wage contract before the field experiment, encourages workers to perform their tasks in a way that is not perfectly aligned with the firm’s preferences (Kapoor , 2010). This means that the interests of the workers are different than the interests of the firm. The interest of the CEO of the firm is to maximize the firm’s profit. While the workers just want to maximize their
10
own utility. The workers get paid based on hourly wages and tips. Hourly wages are fixed, so the only way workers can increase their utility is by increasing the tips. That encourages workers to put great focus on customer service, because the better the customer service, the higher the tips percentage (Kapoor 2010). However, the CEO believes that the best way to increase profits is by increasing customer turnover (Kapoor, 2010, p. 5). Increase in customer turnover would mean that workers should have less focus on customer service. So a conflict of preferences exists. For the experiment Kapoor (2010) designed a contract for the workers which pays performance bonuses to the workers for the number of customers they serve, in addition to their tips and hourly wages. He finds that under the contract of the experiment, the average worker is more productivity, and generates higher profits for the firm (Kapoor, 2010, p. 1). The switch to focus on the customer turnover has a negative impact on the tip percentages, but nevertheless the average workers still earn more. The overall findings by Kapoor (2010) suggest that, giving right incentives to the workers based on the preferences of the firm, has benefits for the workers as well as for the firm.
From the previous studies can be concludes that incentives are a really important instrument for the principals as well as for the agents. With the right incentives, principals can align the preferences of the agents with their own preferences. For example by motivating the agents to increase their productivity, which could lead to an increase in profits, which is beneficial for the principal. Incentives could also lead to an increase in the earnings of the agent, they can collect bonuses if they behave in the way the principal prefers. I saw this in the experiments of Lazear (2000) and Bandiera et al. (2005), they found that the introduction of an incentives pay scheme (piece-rate pay and tournament pay scheme resp.) can lead to an increase of productivity of the employees. Delfgaauw et al. (2013) found that a monetary reward is not always necessary to increase productivity. And Kapoor (2010) found that, giving the right incentives to employees based on the preferences of the firm, has benefits for the employees as well as for the firm. But next to all the positive effects of incentives, I saw that there are also negative effects on productivity due to
incentives. Fehr and Schmidt (2004) showed that if multi-task employees are not monitored in the right way, incentives could lead to a decrease in total output. Because the employees are likely to concentrate their effort only on the task that is rewarded and neglect the other task(s). Gürtler (2010) and Harbring & Irlenbusch (2011) showed that tournament incentives could ensure that employees are encouraged to engage in unauthorized activities. These
11
activities could lead to problems and disadvantages for both employers and employees. My research will be different than the discussed studies. I want to test if tournament incentives increase performance of the agents in the catering industry. I expect that the tournament incentives will have a positive effect on the performance of the employees in my experiment. My hypothesis will be:
H1: The sales of ciabatta’s pesto during the race (with tournament incentives) will be significantly higher (>) than the sales of ciabatta’s pesto without a race (without tournament incentives).
My hypothesis is based on the literature described above. I expect that performance under the treatment of tournament incentives will be higher than without tournament incentives. Because employees in my experiment are working in a small working environment with close friendships. The friendships can enhance the performance of the employees as Bandiera et al. (2009) described. I do not expect that the collusion problem will arise (Gürtler, 2010), because before the tournament started, the employer decided that the reward would be higher if every employee would have personally improved themselves. Therefore employees have no advantages in collusion, because this reduces the potential reward. I expect that there will be a high symbolic value for winning the tournament, because the friendships create a competitive playing field. This will lead to an increase in performance according to Delfgaauw et al. (2013). Besides the effect of the prize the employees can win. Which will also increase the sales on ciabatta’s pesto and thus the performance of the employees, according to Bandiera et al. (2005). I expect that the negative effects of tournament incentives described by Gürtler (2010) and Harbring & Irlenbusch (2011) will not be a problem in my experiment. Because, collusion will not help the employees to win the tournament. There will be only one winner. And it will lead to a lower reward explained before. Excessive risk-taking is not really possible by employees in a restaurant. And at last, I do not expect that there will be a lot of sabotage and mobbing. Because as mentioned before, there are a lot of friendships between the employees, so I expect that they want to win the tournament in a fair way, not by sabotaging. A negative effect that can occur is the problem mentioned by Fehr and Schmidt (2004), namely the multi-task problem. If the employees put all their focus on selling the ciabatta pesto, to win the tournament incentive, the sales of other products could decrease. Also known as the spillover effect. This could
12
lead to an overall decrease in sales, which is not beneficial for the employer. I will discuss this problem later on.
Methodology
In this section I will describe the data I will use and how the data was collected. Furthermore I will describe the research design and the empirical models I will test.
As mentioned before, this research consists of an own field experiment at a real restaurant. It’s a medium sized Italian restaurant with 31 employees working as
waiter/waitress. To answer my research question: ‘’Do incentives in tournaments have a positive effect on the performance of employees in the catering industry?’’, I have organized a tournament for all waiters/waitresses. Unfortunately it was not possible to create a control group. First, because everyone is participating. The employees know each other well, it would be strange if not everyone can join the tournament. The second reason is that the control group already knows how the tournament goes into its operation. A control group should not hear anything from the experiment, otherwise it would affect the results.
The focus of the experiment is on the sales of ciabatta’s pesto. The reason of this is because it’s difficult to measure the overall performance of employees in the catering sector, because they have a lot of tasks that they have to perform. Because of this, I have decided to measure the sales of one product, and see if employees would increase their sales in this product if there is an incentive at stake. I have chosen to measure the sales of the ciabatta pesto. A ciabatta pesto is just some sort of bread that you can order as a little bite before dinner, like an appetizer. I have chosen for this product because I think that the employees have the most influence to sell these ciabatta’s to customers. Often these ciabatta’s are ordered before the customers have even looked at the menu. I think this can bring some negative spillover effects. But I will discuss that later in this section.
The tournament was held from 1 to 24 December. After the 24th of December, the
restaurant was closed for 2 days due to Christmas, that’s why the tournament ended on the 24th. The employee with the highest average ciabatta’s sold per hour, wins the tournament.
There were several rules according to the tournament:
- The tournament only applies in the evenings. It starts at 4.30 pm, when most evening shifts begin. And it ends at the time the kitchen closes. The kitchen closes at 8 pm on
13
Sunday till Wednesday and at 9 pm on Thursday till Saturday. During this time the sale of the ciabatta pesto is tracked per employee.
- Only the employees who take the orders from the customers participate. So not the employees behind the bar, the baristas. There is no permanent employee behind the bar in this restaurant. Each time the employees rotate between working behind the bar and taking orders/serving food to the customers.
- The employees write down the amount of ciabatta’s pesto they have sold by themselves each evening during the tournament.
- The amount of ciabatta’s sold will be divided by the amount of working hours between 4.30 and 8/9pm. This way I can get an average per employee. This means that full-time employees have no advantage during this tournament.
- There will also be a factor assigned to the amount of sales for each evening. So a ciabatta sold on a quiet evening, is worth more than a ciabatta sold on a very busy evening. Because in the last case the employee will have more chances to sell one. The factor is determined as follows: There are 3 categories of an evening. Quiet, average and busy. During the tournament there will be a total of 24 evenings. The 8 lowest sales will be placed in the category Quiet. The 8 highest sales will be placed in the category Busy. And the last 8 sales, which are between quiet and busy, will be placed in the category Average.
Table 1
There is another factor for each category, which you can see in table 1. A short
example to see how these factors work. Assume worker 1 sells 6 ciabatta’s pesto on a busy evening and 2 on a quiet evening. In total worker 1 sold: (6*0.9)+(2*1.1)= 7.6 ciabatta’s pesto.
At the end of the tournament, the employee with the highest average ciabatta’s pesto sold per hour worked, wins the tournament. There will be only 1 winner. The winner receives a prize. Which is, an evening of free cocktails for 2 persons. The reward was mentioned before the tournament started. Besides that the employer decided that the reward would be higher
Quiet Average Busy
14
if every employee would have personally improved themselves, the employees know this. Before the tournament started, the sales of ciabatta’s pesto per individual was counted in the whole month November. The employees did not know that their sales of ciabatta’s pesto were tracked in November. I have calculated the average ciabatta’s pesto sold per hour for every employee in November. I can compare this data to the average ciabatta’s pesto sold per hour in the month December. In order to determine whether the data in December is significantly different from the data in November, I use a
non-parametric test because of the small number of observations. The test I will use will be the Wilcoxon signed rank test. This test is used to compare a continuous variable between two paired groups. My data meets these characteristics because I use exactly the same group of employees in both November and December.
Further I will perform an OLS regression. I will include 3 dummy variables and test if they are significant or not. The regression equation will be as followed:
𝑌𝑖 =∝ +𝛽1 ∗ 𝑇𝑜𝑢𝑟𝑛𝑎𝑚𝑒𝑛𝑡 + 𝛽2 ∗ 𝐸𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒 + 𝛽3 ∗ 𝐸𝑥𝑡𝑟𝑜𝑣𝑒𝑟𝑡
+ 𝑇𝑜𝑢𝑟𝑛𝑎𝑚𝑒𝑛𝑡 ∗ 𝐸𝑥𝑡𝑟𝑜𝑣𝑒𝑟𝑡 + 𝜀𝑖
Yi is the dependent variable, and that is the total average in the months of November and December. The alpha is the constant in this model. Tournament, Experience and Extrovert all are dummy variables. Tournament stands for sales during the tournament from 1 to 24th of
December. Experience is a dummy variable which takes on value 1 if an employee is working in this restaurant for more than a year, and 0 if otherwise. I have chosen this criterion because after 1 year an employee has experienced all possible situations in the restaurant, so they can consider themselves as experienced. In total 16 of the 31 employees are experienced. Previous research has shown that there is a positive correlation between working experience and job performance (Quiñones, Ford, & Teachout, 1995). I expect that this is also the case in my research, so more experienced employees will have a higher performance in this tournament. The variable extrovert is also a dummy variable. This variable will be included for every employee who has an extrovert personality. I determined the personalities of the employees by a short survey among all employees, which they filled in themselves. The questions where:
15
1. Would you say you have many or few conversations with customers in the restaurant, compared to colleagues?
A) Many conversations B) A few conversations
2. Are you more of a listener or more of a talker?
A) Talker B) Listener
3. How do you feel about being alone?
A) I like it, I need some time alone B) I don't like it at all
4. What is the best way to describe the kind of friends you have?
A) I have a lot of acquaintances at a lot of different places and a few real friends. B) I have a few very close friends
5. How would you describe yourself?
A) Calm and self-contained B) Enthusiastic and outgoing
The questions where based on a survey made by Wagenaar, D. (2017). Due to this short survey I could place the employees in 2 categories: Extrovert personality or Introvert personality. The an employee filled 3 or more ‘extrovert’ answers I considered them as extrovert. If otherwise, I considered them as introvert. In total 18 of the 31 employees have an extrovert personality.
The last variable in an interaction variable between tournament and extrovert. With this variable I can test if extrovert employees perform better during the tournament than introvert people. Subsequently I can examine with the OLS regression if the dummy variables are significant. The final factor of the regression equation is the error term (𝜀𝑖). This factor
16
Hypotheses
Several hypotheses can be derived for this research. My main hypothesis is:
H1: The sales of ciabatta’s pesto during the race (with tournament incentives) will be significantly higher (>) than the sales of ciabatta’s pesto without a race (without tournament incentives).
This hypothesis is based on the literature of Bandiera et al. (2005) and Delfgaauw et al. (2013). These studies show that the introduction of a tournament pay structure can lead to an increase in productivity of the employees. And that the tournament pay structure can also enhance the productivity through a symbolic effect explained earlier.
I will test this hypothesis through a Wilcoxon signed rank test as mentioned before. I will do an one-tailed test at a significance level of 5% to see if the average ciabatta’s pesto sold during the tournament is significantly higher than the average ciabatta’s pesto sold in November. But I will also run an OLS regression where I will test the dummy variable
tournament against the total average sales of ciabatta’s pesto. Furthermore I will include the other dummy variables in this OLS regression, to see if these have an effect on the
tournament dummy.
Besides the main hypothesis I also have three other sub hypotheses. With those hypotheses I will investigate some other factors next to the research question. Which may offer some added value to this research. The first sub-hypothesis is:
The second sub-hypothesis is:
H2: The sales of ciabatta’s pesto by extrovert people is significantly higher (>) than the sales of introvert people.
I expect that extrovert employees will have a significantly higher sales ciabatta’s pesto than introvert employees. Research has shown that extrovert people score higher on
performance than introvert people (Chiang, Hsu, & Shih, 2014). I expect that is also the case in my research, because extrovert people are normally more outgoing and less afraid to approach people and thus customers. So they have more conversations with customers. Because of this I expect that extrovert employees are better in their ‘sales pitch’ than introvert employees. As a result of this, I expect that customers are more inclined to order a
17
ciabatta pesto from extrovert employees. And thus the sales of extrovert employees will be higher than introvert employees. This hypothesis will be tested within the OLS regression. As mentioned before, I will include a dummy for extrovert employees. If the dummy is
significant, it means that extrovert employees in this restaurant sell more ciabatta’s pesto than introvert employees. I also expect that extrovert employees will perform better than introvert employees, during the tournament, I will test this through the interaction variable I added in the regression equation.
The third sub-hypothesis is:
H3: The sales of ciabatta’s pesto of employees who are more experienced/have more abilities is significantly higher (>) than the sales of ciabatta’s pesto of employees who are less experienced/have less abilities.
This hypothesis is based on the study of Quiñones et al. (1995). They show that employees with more working experience on average have a higher productivity then employees with less working experience. I expect that this will be the same in my research. I will test this hypothesis through a OLS regression. I have added a dummy variable which takes on value 1 if an employee is working in this restaurant for more than a year, and 0 if otherwise. I have chosen this criterion because after 1 year an employee has experienced all possible
situations in the restaurant, so they can consider themselves as experienced. In total 16 of the 31 employees are experienced.
The fourth sub-hypothesis is:
H4: The sales of desserts during the race is significantly lower (<) than the sales of desserts without a race.
This hypothesis is based on the literature of Fehr and Schmidt (2004). Also called the multi-task problem. Fehr and Schmidt (2004) state that if employees put all their focus on selling one product, to win the tournament incentive, the sales of other products could decrease. Also known as the spillover effect. This could lead to an overall decrease in sales, which is not preferable for the employer. I have discussed this problem before. I have chosen for the spillover effects on desserts, because I think that the employees have also a lot of influence to sell desserts to customers. There are some homemade desserts on the menu, if
18
that the sales of desserts is not significantly higher. Because due to the literature, I expect that the employees focus more on the sales pitch of the ciabatta pesto and forget about the desserts. I will test this hypothesis by comparing the total sales of desserts in November with the total sales of desserts in December during the race, and correct these for the total sales in these months.
The last sub-hypothesis is:
H5: The sales of ciabatta’s pesto during the race is not significantly different (≠) than the sales of ciabatta’s pesto after the race.
With this hypothesis I want to investigate whether the incentives only work in the short run, or also in the long run. I expect that the sales of the ciabatta’s pesto after the race is still significantly higher than the sales before the race started, because the employees still have the tournament in mind. I expect that it is still in everyone’s routine to try to sell as much ciabatta’s pesto as possible. Besides that, the winner of the tournament has been
announced and the employer knows all the averages of every individual. I think that most employees want to improve themselves to impress the employer. Mainly the employees who are less experienced. I will test this hypothesis by means of the Wilcoxon signed rank test. I will do a two-tailed test with a significance level of 5%. With this test I can calculate if the average ciabatta’s pesto sold after the race does not significantly differ from the
ciabatta’s pesto sold during the race. As soon as the research is finished, the performance of the employee is back at is old level (before the research).
Results
First I will consider the average ciabatta’s pesto sold in control and treatment. The result show that the experimental treatment had the expected effect on the average ciabatta’s pesto sold per hour. 18 of the 31 workers managed to increase their average sales of ciabatta’s pesto during the tournament. The total average has increased from 0.54 to 0.75 ciabatta’s per hour, shown in table 2. The increase in average sales during the tournament is significantly higher than the average sales in the control month November. The result is significant at the p < .05 level for an one-tailed test. I have used an one-tailed test because I am only testing for the possibility in one direction. Namely if the average sales of ciabatta’s pesto is ‘higher’. I have corrected the averages for the difference is sales in November and
19
December during the tournament. I did this with different factors for different sales per day, like I have explained before. I have also considered to base the factors on the total amount of sales in the in months November and December and then determine three groups. In this way I came to the same result, so I decided to just chose one of the two ways.
Table 2 Dependent variable Control average (November) Treatment average (During tournament) Control < Treatment (p-value)
Total average sales of ciabatta’s pesto
0.54 0.75 0.034*
Average sales €2132.9 €2273.14 0.9734
Note: To test the level of significance, the Wilcoxon signed rank test is used, * = sign. for a 5% level
Table 2 also shows the average sales per day. The sales in December are only
€2273.14−€2132.9
€2132.9 ∗ 100% = 6.61% higher in December, which is not significant. That’s
showed in table 2.
From this results I can conclude the incentives given in the tournament structure leads to a significantly higher performance on the sales of ciabatta’s pesto of the employees in this restaurant. This is what I expected before the tournament started. This shows that hypothesis 1 follows from the data:
H1: The sales of ciabatta’s pesto during the race (with tournament incentives) will be significantly higher (>) than the sales of ciabatta’s pesto without a race (without tournament incentives).
This confirms the theory of Bandiera et al. (2005). The tournament pay structure has led to a higher performance of the employees. The theory of Delfgaauw et al. (2013) could also have played a role in the increase in performance. Namely that the increase in performance is caused by a high symbolic value of the tournament. For now I cannot say much about that, because I do not have the data for this. This could be an interesting topic for further research.
20
In the second result I will discuss the effect of all dummy factors on the performance of employees during the tournament, which I used in my OLS regression. First I have considered the collinearity of the independent variables. Mainly between the dummy variables extrovert and experience. These two could be highly correlated in the catering industry. This is because often (but not always), the employees that last the longest, so with the most working experience, are extroverted people. If these two variables are strongly correlated, this can lead to multicollinearity. Multicollinearity affects the calculation of the coefficients, because the independent variables can overlap. This reduces the reliability of the coefficients and the variables.
In this case there is no multicollinearity. Because the dummies extrovert and
experience are only slightly correlated. The correlation between extrovert and experience = 0.077. Because of this I do not have to worry about multicollinearity. The weak correlation could be caused by the fact that there are more extrovert employees than experienced employees. Another reason could be that new unexperienced employees are extrovert.
Furthermore, the standard errors in this regression are robust. This means that the standard errors stay consistent if there is any form of heteroscedasticity.
21
In the first regression shown in table 3, I see that the dummy variable tournament is significant again. This is in line with the non-parametric test shown in table 2. It shows that the tournament leads to an positive effect of the performance of the employees on the sales of ciabatta’s pesto. The dummy is significant at the p < .05 level for a two-tailed test with a 5% significance level.
In the second regression I added the dummy variables extrovert and experience. The dummy variable tournament is also significant is this regression. Again for a 5% significance level. The dummy variable extrovert is significant as well. This means that the dummy variable extrovert has no impact on the variable tournament. This is what I expected. The result is significant at the p<.01 level for a two-tailed test. This implies that my second sub-hypothesis is true:
H2: The sales of ciabatta’s pesto by extravert people is significantly higher (>) than the sales of introvert people.
The sales of ciabatta’s pesto is significantly higher amongst extrovert people. This result confirms the theory of Chiang et al. (2014). They have shown in their research that extrovert people score higher on performance than introvert people.
The dummy variable experience is not significant at a 5% significance level. This is not what I expected. I expected that experienced employees will have a higher performance is this tournament than less experienced employees. Based on the previous research by Quiñones et al. (1995). They have shown that there is a positive correlation between working
experience and job performance, which I have discussed before. A higher level of working experience would lead to higher job performance. This is not the case in my research. A reason for this could be that less experienced employees, were extra motivated to sell ciabatta’s pesto. Because they still had to prove themselves or they wanted to impress the employer. This means that my third sub-hypothesis is rejected:
H3: The sales of ciabatta’s pesto of employees who are more experienced/have more abilities is significantly higher (>) than the sales of ciabatta’s pesto of employees who are less experienced/have less abilities.
22
In the third regression, the interaction variable Tournament x Extrovert is added, to test if extrovert employees perform better during the tournament than introvert employees. Table 3 shows that the dummy variable tournament is still significant (p=0.029). This implies that the tournament incentives lead to a significant higher performance on the sales of ciabatta’s pesto for introvert employees. The interaction variable is also significant (p=0.001). This implies that the tournament incentives also lead to a significant higher performance on the sales of ciabatta’s pesto for extrovert employees (Tournament + Tournament x Extrovert), because both dummy variables are significant. From this results I can conclude that both introvert and extrovert employees increase in performance during the tournament. My expectation that only extrovert employees will perform better during the tournament is rejected. They both increase in performance.
My fourth sub-hypothesis is:
H4: The sales of desserts during the race is significantly lower than the sales of desserts without a race.
I expected that the tournament incentives would lead to some spillover effects on other products, because the employees are more focused on selling ciabatta’s pesto instead of other products. To test this hypothesis I compared the sales of desserts in November with the sales of desserts in December. The total amount of desserts sold in November was 509, and the total amount of desserts sold in December was 504. The amount of desserts sold has decreased a little bit. But it’s not significantly lower at a 5% significance level. So my
hypothesis is rejected. I have corrected these numbers for the sales. The difference in sales in the month November and December are showed in table 2. The average sales in
December is 7% higher than in November. So I expected that the total amount of desserts should also increase with about 7%. Then the desserts sold in December should be around 544, but this is not the case. My suggestion is that the employees have indeed put more focus on selling ciabatta’s pesto and that this has led to the small decrease in amount desserts. This suggestion is based on the literature of Fehr and Schmidt (2004). They state that if employees put all their focus on selling one product, to win the tournament incentive, the sales of other products could decrease. In this case it is not significant, but it could be a reason for the decrease of desserts.
23
The last result also has the expected effect. In my fifth sub-hypothesis I expected that the average sales of ciabatta’s pesto after the tournament was not significantly different from the average sales during the tournament:
H5: The sales of ciabatta’s pesto during the race is not significantly different (≠) than the sales of ciabatta’s pesto after the race.
With this hypothesis I wanted to test if the reaction to incentives only works in the short term or also after the incentive were abolished, in the ’long term’. Table 3 shows that the average sales of ciabatta’s pesto has decreased from 0.75 to 0.70. This is a small decrease, and it is therefore not significant (p=0.2775). This means that the average sales of ciabatta’s pesto after the race is not significantly different from the average sales of ciabatta’s pesto during the race. This time I used a two-tailed test, because I now wanted to allow for the possibility that change goes in both directions.
Table 4 Dependent variable Treatment average (During tournament) After treatment average (After tournament) Treatment = After treatment (p-value) Average sales ciabatta
pesto
0.75 0.70 0.2775
Note: To test the level of significance, the Wilcoxon signed rank test is used, * = sign. for a 5% level
Conclusion & Discussion
In this paper, I designed and conducted a field experiment at a medium sized restaurant to study the effect of tournament incentives on the performance of employees in the catering industry. I also investigated if the effect of these incentives only hold in the short term, or also in long term. Furthermore, I investigated whether certain characteristics of employees are dominant for a higher performance and I investigated whether the incentives also had negative effects.
I have found that the average performance has increased during the tournament. Measured through the sales of ciabatta’s pesto per individual employee. I have also found that the performance of the employees was still high after the tournament was finished. The effect of the incentives did not disappear immediately when the tournament ended. So the incentives also worked after the experiment abolished. These findings correspond to the
24
findings of the literature I have discussed. From the previous researches I can conclude that incentives are a really important instrument for the principals as well as for the agents. With the right incentives, principals can align the preferences of the agents with their own
preferences. Incentives could also lead to an increase in the earnings of the agent, they can collect bonuses if they behave in the way the principal prefers. I saw this in the experiments of Lazear (2000) and Bandiera et al. (2005), they found that the introduction of an incentives pay scheme (piece-rate pay and tournament pay scheme resp.) can lead to an increase of productivity of the employees. Delfgaauw et al. (2013) found that tournaments with and without monetary rewards both can be used to increase productivity. And Kapoor (2010) found that, giving the right incentives to employees based on the preferences of the firm, has benefits for the employees as well as for the firm.
Furthermore, I found that extrovert employees performed significantly higher than introvert employees under the tournament pay structure, this is in line with the literature of Chiang et al. (2014). Employees with more working experience did not perform better than employees with less working experience. A reason for this could be that less experienced employees, were extra motivated to sell ciabatta’s pesto. Because they still had to prove
themselves or they wanted to impress the employer.
The tournament incentives have also brought negative effects, namely spillover effect. This phenomenon, also called the multi-task problem,has already been raised by Fehr and Schmidt (2014). They state that if employees put all their focus on selling one product, to win the tournament incentive, the sales of other products could decrease. Although the spillover effect on desserts was not significant, there was a decrease in the amount of desserts during the tournament compared to the control month November, while sales had increased. My suggestion is that the employees have put more focus on selling ciabatta’s pesto and that this has led to a small decrease in amount desserts sold.
The results show that the average ciabatta’s pesto sold during tournament incentives are significantly higher than without tournament incentives, but how valuable are these results and are there some new insights for future research? First of all, there are some limitations in my research. The most important one is the sample size, only 31 employees participated in my research. This was driven by the number of employees working at the restaurant where I conducted my research. For a larger sample size, a large-scale restaurant should be used. But due to the short time of my research this was not feasible for me.
25
Extending research to large-scale restaurant could be valuable.
Another problem due to the small sample size, is that I could not make a control group. Therefore the control group I have used is the performance of the employees in the month November. Because the sample size was too small to make another control group. The second reason is that the control group already knows how the tournament goes into its operation. A control group should not hear anything from the experiment, otherwise it would affect the results.
A limitation due to the short time of the research is that I could not test hypothesis 5 properly. Hypothesis 2 states: H5: The sales of ciabatta’s pesto during the race is not
significantly different (≠) than the sales of ciabatta’s pesto after the race. I have found that
the average sales of ciabatta’s pesto did not significantly decrease the week after the tournament has ended. But I cannot say anything about the long term effect of the
incentives. Because I did not have the tools and the time to measure the sales of ciabatta’s pesto per individual after the tournament for longer than 1½ week.
An interesting point for future research could be the effect of the symbolic value of winning a tournament on the performance per individual employee. In this case I did not have the data to say something about the symbolic value. Delfgaauw et al. (2013) have already shown that the tournament structure leads to higher performance, even if there is no monetary reward at stake. They designed and conducted a field experiment with two tournaments, in one tournament there were monetary rewards at stake. In the other tournament there were no monetary rewards. From the results they found that in both tournaments the average productivity increased. A new insight could be, what the effect of the symbolic value of the tournament pay structure has, on different employees with different personalities.
26
References
Bandiera, O., Barankay, I., & Rasul, I. (2013). Team incentives: evidence from a firm level.
Journal of the European Economic Association, 11 (5). pp. 1079-1114. ISSN 1542-
4766.
Bull, C., Schotter, A., & Weigelt, K. (1987): Tournaments and Piece Rates: An Experimental Study, Journal of Political Economy, 95(1), 1-33.
Chiang, Y., Hsu, C., & Shih, H. (2014). Experienced high performance work system,
extroversion personality, and creativity performance. Asia Pac J Manag, Vol. 32, pp. 531–549.
Delfgaauw, J., Dur, R., Sol, J., & Verbeke, W. (2013). Tournament Incentives in the Field: Gender Differences in the Workplace. Journal of Labor Economics, Vol. 31, No. 2, pp. 305-326.
Ederer, F. (2010).Feedback and Motivation in Dynamic Tournaments. Journal of Economics
& Management Strategy, Vol.19(3), pp.733-769.
Fehr, E., & Schmidt, K.M. (2004). Fairness and incentives in a multi-task principal-agent model. Scandinavian Journal of Economics, 106(3), 453–474.
Gibbons, R. (1998). Incentives in Organizations. Journal of Economic Perspectives, Vol. 12, no. 4, pp. 115-132.
Gneezy, U., Meier, S., & Rey-Biel, P. (2011). When and Why Incentives (Don’t) Work to Modify Behavior. Journal of Economic Perspectives, Vol. 25, No. 4, pp. 191–210. Gürtler, O.(2010): Collusion in homogeneous and heterogeneous tournaments, Journal of
Economics, 100(3), 265-280.
Harbring, C. & Irlenbusch, B. (2011). Sabotage in Tournaments: Evidence from a Laboratory Experiment, Management Science, 57(4), 611–627.
Kapoor, S. (2010). Incentive Provision in Multitask Jobs: Experimental Evidence from the Workplace. University of Toronto, Department of Economics.
Lazear, E.P. & Rosen, S. (1981). Rank-Order Tournaments as Optimum Labor Contracts.
Journal of Political Economy, Vol. 89, No. 5, pp. 841-864.
Lazear, E.P. (2000). Performance pay and productivity. American Economic Review, 90(5):1346– 1361.
Quiñones, M.A., J.K., Ford, & M.S., Teachout. (1995). The relationship between work experience and job performance: a conceptual and meta-analytic review. Personnel
Psychology, 00315826, Winter95, Vol. 48, Issue 4, pp. 887–910.
Wagenaar, D. (2017). Are You An Introvert Or Extrovert Quiz? Retrieved from
https://www.proprofs.com/quiz-school/story.php?title=introversion-vs-extroversion-mbti-preference-1.
