The effect of a quota on individual performance

(1)

The effect of a quota on individual performance

Gender diversity in boards can lead to better firm performance. Since women are underrepresented in boards, it is important to employ more women in high power positions. A way to employ more women is via an employment quota. Many studies investigated the effect of a quota on entry decisions in tournaments. This paper however investigates whether a quota system has an effect on individual performance or not. The results do not show any significant effects of better or worse performance under a quota system.

Name: M. M. Korstenbroek Student number: 10646612

Thesis supervisor: dr. S. Dominguez Martinez BSc- programme: Economics & Business Specialization: Finance & Organization

(2)

Statement of Originality

This document is written by Maud Korstenbroek 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 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)

1. Introduction

If boards are diverse in gender, it has a positive effect on the performance of the board (Sabatier, 2015). According to Sabatier’s study, gender diversity has a positive effect on the economic performance and corporate social responsibility in the short run. Sabatier did not study the long run, since there is only data on very diverse boards since the employment quotas are implemented (2015). According to a report from the European commission, more gender diverse boards also have a positive effect on corporate governance, team performance and the quality of decision making (Dorrough et al., 2016). Therefore, it seems to be in the best interest of all companies to have a more balanced sex ratio in the boards. The problem is that in most boards, females are underrepresented. In the three European countries with the largest diversity in the boards, one out of three board directors is female. But almost never one of the female directors is leading the board or possesses the CEO position (Akyol et al., 2015). Most of these high in power positions are male-dominated (Niederle and Vesterlund, 2007). On contrary, in Japan only 1% of the board directors is female (Akyol et al., 2015). On reason is that women have the tendency not to choose for these male-dominated jobs (Heilman and Mardenfeld, 1984) and instead choose less demanding jobs. Females do not go for jobs like executives or board members (Dorrough et al., 2016). Nevertheless, the percentage of women in the workforce is 56% (Sabatier, 2015) and there is a high increase of women in professional schools (Heilman and Mardenfeld, 1984). To make sure men and women are equally represented among all positions in all fields and increase firm performance, an employment quota could be the solution.

In Europe, the attention to employment quotas emerged since Norway introduced an employment quota in 2003 (Akyol et al., 2015). In the countries Iceland, Spain and France a quota is forcing the boards to have at least 40% women in the boards by 2013, 2015 and 2017 respectively (Akyol et al., 2015). The quota is not the same for all countries. For example, the quota in Norway is 40% women, while the forced quota for Germany is 30% women (Akyol et al., 2015). First the quota regulations were mainly used for political participation, now the quotas are also largely used in other sectors (Dorrough et al., 2016).

Many literature investigates the differences between men and women and their behaviour against a quota system. People have different reactions to a quota system. This

(5)

leads to the research question of this paper. Does a quota have an effect on individuals’ performance?

In this paper I will investigate whether a quota has an effect on performance or not. I designed an experiment where I want to find out if participants work harder, less hard or work the same when they know that they will compete in a test where we use a quota in favour of women to select the final winners. I collected the data via my own online experiment in which participants had to perform two mathematical tests. Results show that the performance of men and women is significantly the same. Other results are that a quota has an effect on performance, and that women in quota groups perform better under a quota treatment. Unfortunately, these results in this study are not significant.

In most of the literature, the effect of a quota on entry decisions in tournaments is investigated. In my research, I investigate whether a quota has an effect while competing in a tournament. Participants did not have the decision to enter. In this paper it is investigated, whether the quota itself has an impact on the performance of the participants. Therefore, this is research is new. The outcome of this research is relevant. The reason can be that if firms are selecting new board members or hiring new employees under an employment quota, it can be beneficial to the firm to know if the participants perform different than they would without a quota. In order for participants to behave different, they have to know about the quota.

For the remainder of this thesis, an overview of the related literature is showed in section two. This literature review provides theoretical information in order to better understand the topic. In the third section, the method of the research is discussed. It provides information about the data, the experiment and the models used. In the fourth section, the results are shown. The results will be discussed in the fifth section. Finally, in the sixth section, an overall conclusion is provided.

(6)

2. Theoretical framework

2.1. Differences between gender

Before discussing the effect of quotas on performance, it is important to take a look at the differences between men and women. There are two important differences that will be discussed. In this section, we look at the differences between men and women for competition and in risk taking.

2.1.1. Competition

Competitive environments work better for men than for women, even though the qualities of men and women are equal. The performance of men increases when they work in competitive environments. On the other hand, women tend to stay out of competition environments at all (Balafoutas and Sutter, 2012). When men have to work in a group and the group has to compete to other groups, men add more to the group than when the groups did not interact. This effect was not found for female cooperation (Powell and Ansic, 1997). Because of the differences in behaviour towards competition between men and women, women have less opportunities to get a promotion, and the wages of females may be smaller than the wages of males (Balafoutas and Sutter, 2012).

As stated before, men perform better under competition. This is shown to be true in the experiment of Balafoutas and Sutter (2012), since the performance increased in the second stage (with tournament) compared to the first stage (with piece-rate). Nevertheless, part of this increase in performance can also be caused by the fact that participants learn and practice from the first stage (Balafoutas and Sutter, 2012). Balafoutas and Sutter (2012) found that the men in the experiment, choose to go for the tournament twice as often as women do. This is consistent with the literature.

2.1.2. Risk aversion

if there is a difference in risk aversion between gender, it is very important because it can explain many economic findings (Charness and Gneezy, 2011). Many literature states that women are more risk averse than men are (Borghans et al., 2009). Still the results of studies in risk aversion are not always in line with each other (Powell and Ansic, 1997). Interesting is, that men and women have different strategies when they have to make financial decisions.

(7)

Nevertheless, the strategy does not have a significant effect on the performance (Powell and Ansic, 1997). It does not matter whether the investment, project or test is familiar to the participants, women still are more risk averse than men are (Powell and Ansic, 1997). A lot of research is done on the risk averseness of men and women in financial situations. Charness and Gneezy (2011) investigated with an experiment if there is a difference between gender in risk aversion. They found that for risky investments, women invest a smaller amount than men do. Therefore, they can conclude that females are more risk averse in the financial situations.

2.2. Problems with quotas

The reason that firms have to apply quota systems and use preferential selection, is that women are still underrepresented in the public and private sector. Nevertheless, a quota could have some psychological consequences. Most results of the psychological studies below are found by creating hypothetical situations. Participants who are selected for a position because of preferential selection on gender, believe that other people working in the same company have negative expectations about their skills (Barocas Alcott and Heilman, 2001). The quotas are to increase the number of women in the company. As a consequence, women have negative expectations about their own skills. It does not matter how qualified the woman is for the job. The woman will still believe that others have negative expectations if she was selected because of the preferential selection (Barocas Alcott and Heilman, 2001). According to Dorrough et al. (2016), females who believed they were selected pure on the basis of preferential selection evaluate their own leadership skills more poorly. The females also have the tendency to choose less demanding tasks. These effects were not found when women thought they acquired the job only because of performance (Dorrough et al., 2016).

Heilman and Mardenfeld Herlihy (1984) did research on participants’ job interests if females got their job because of merit or because of preferential selection based on gender. Heilman and Mardenfeld Herlihy (1984) created an experiment where they asked participants about their job interests for a hypothetical job. They found that women who believe getting the job because of preferential have less interest in the job than they would have if it was because of their performance. For males, the interest in the job was also lower if they believed that females got the job because of preferential selection. Thus, when

(8)

females got their jobs because of their performance, then more balanced gender ratios have a positive effect on job interests for both men and women (Heilman and Mardenfeld Herlihy, 1984). The way woman acquires a job, has a big impact on the interests she has in the job (Heilman and Mardenfeld, 1984).

Quotas or intervention policies, which make it easier for females to win tournaments, can have a negative effect (Balafoutas and Sutter, 2012). In order to let a woman win the tournament, with the help of the policy, better performing men can be passed by. But in contrary, the intervention policies can have as effect that more high-performing women enter the tournament. Without the intervention policy, these high performing women would have chosen the piece rate (Balafoutas and Sutter, 2012). Interventions policies can make sure that the group of total participants in the tournament is of better quality than it would be without the intervention policy (Balafoutas and Sutter, 2012). A quota therefore does not have efficiency losses.

2.3. Empirical research Niederle and Vesterlund

Niederle and Vesterlund (2007) did research on the entry decisions of women and men. They look at piece-rate games and tournament games. They designed an experiment with multiple tasks. Each group consisting of six participants, three of them male and three females. In the piece rate task, they let participants do a simple math test, where the participant will be rewarded with $0,50 on the number of correct answers. In the tournament task, the math test is similar to the other test, but now the two best performing participants will be rewarded with $1,50 per correct answer. According to Balafoutas and Sutter (2012), men on average perform better than women, though the difference in performance is not significant. On the other hand, Niederle and Vesterlund (2007) found that there was no significant difference in the performance in these math tests between men and women. Under the tournament, both men and women perform significantly better than they do under the piece rate (Niederle and Vesterlund, 2007). Nevertheless, women prefer the piece-rate over the tournament. Niederle and Vesterlund give possible explanations for this difference. One is that women are likely to be more risk averse than men are (Niederle and Vesterlund, 2007). Another difference between genders that was provided, is that women dislike receiving feedback on their relative performance more than men do (Niederle and Vesterlund, 2007). Since the relative performance is important in a

(9)

tournament, it makes sense that women choose for the piece rate instead. Furthermore, women are less optimistic than men when it concerns their beliefs (Niederle and Vesterlund, 2007). This can also be one of the reasons why women prefer to choose for the piece rate. It is important to keep in mind that these gender differences exist when taking conclusions about the behaviour of men and women in entry decisions.

2.4. Empirical research Niederle et al. on quotas

After the research of Niederle and Vesterlund in 2007, Niederle et al. (2013) did more research on the entry decisions between genders. In this new research, they also take quotas into account. The reason why quotas are implemented, is that the group of applicants that can be selected as winners is more diverse. The quota says that among the two winners out of the six participants, at least one has to be female (Niederle et al., 2013). This means that as a woman, you either have to be the best performing participant, or you have to be the best performing woman. For a man this means you have to be both the best performing men and the best performing participant. This quota means that the expected payoffs for females will be higher, while the expected payoffs for males will be lower (Niederle et al., 2013). Same as in the previous experiment, Niederle et al. (2013) found that the performance of both men and women increases in the tournament. Because of the quota, more women choose to compete in the tournament instead of the piece-rate (Niederle et al., 2013). This is a very interesting finding of this research, because it implies that women prefer to compete against women. When there was no quota, to win you have to be one of the two best performers. Now with quota, as female, you only have to be the best female in order to win the tournament. This causes that de group of entrants is much more diverse than without quota (Niederle et al., 2013). Niederle et al. (2013) investigated if no better performing men will be passed by in order to let a less qualified woman win the tournament. They found that under a quota system, more qualified women enter the tournament than without the quota. Therefore, the selection of participants that can be selected as winners is bigger. As a result, no better performing men will be passed by since the selection also contains better performing women under the quota (Niederle et al., 2013). Akyol et al. (2015) shows that females have negative feelings when they were selected on the basis of preferential selection. Other studies show, that females are more likely to enter competitive situations when there is a quota, because the probability of

(10)

winning is higher since they only have to compete against other women (Niederle et al., 2013; Balafoutas and Sutter, 2012). For males, the performance does not change under an employment quota for women (Niederle et al., 2013; Balafoutas and Sutter, 2012), but their interests in a job is lower if a female acquired the job because of preferential selection (Heilman and Mardenfeld Herlihy, 1984).

2.5. Effect of quotas in tournament entry decisions

Niederle et al. (2013) find in their experiment that more women enter the tournament when a quota is implied. This in line with the findings of Balafoutas and Sutter (2012). The increase of participants is the strongest for policies where women were given extra units of answers before the game started (Balafoutas and Sutter, 2012). A repetition of the game, leads to a positive increase in participants, but this increase is not significant.

For male participants, the intervention policies have no consequences on the entry decision (Niederle et al., 2013). Therefore, according to Balafoutas and Sutter (2012), a quota system only affects the entry decisions of women, and not of men.

The conclusion of Balafoutas and Sutter (2012) their experiment is that strong intervention policies are the best way to attract high performing women to the tournament.

2.6. Hypotheses

To answer the research question, three hypotheses will be tested. The first hypothesis is: There will be no large differences in the math performance between men and women. According to Niederle and Vesterlund (2007) there are no differences in performance between men and women. In this study’s experiment, the participants have to perform a similar mathematical tasks as in the experiment of Niederle and Vesterlund (2007). Therefore, I expect to find no large differences in the performance between men and women.

The main focus of this paper will be on the second hypothesis. The second hypothesis is: A quota treatment has an effect on participant’s performance. According to Balafoutas and Sutter (2012), more women enter tournaments when there is a strong quota system. Niederle et al. (2013) say that this is because women prefer competing against women. For both men and women, performance increased under the quota system in the tournament. In contrary to Niederle et al. (2013), according to the theory literature, quotas can have

(11)

negative psychological effects for men and women (Brett et al., 1991). These effects can also influence the performances. Therefore, the effect of a quota system can be positive or negative.

Since the reactions of men and women towards a quota can be different, I will investigate if men and women differ in their performance under a quota. This is the third hypothesis of this paper.

(12)

3. Method

3.1. Research setting

For this research, I designed my own experiment. Using a program called Qualtrics I was able to design this experiment. In this experiment, participants have to perform two similar mathematical tasks. Each participant has to sum up five 2-digit numbers for a time period of three and a halve minute. Participants are not allowed to use a calculator, but the participants are aware of the fact that they are allowed to use scratch paper. All the numbers were randomly drawn, presented this way:

33 + 27 + 11 + 40 + 83 =

After each question, there is a blank space left, to fill in the answer. All the questions of one test are placed on the same page. So the participants are able to see all the questions.

At the beginning of the experiment, participants are told that they are participating in a hypothetical job interview. The participants are aware of the fact that it is a hypothetical, and not a real job interview. Trying to motivate the participants, a gift card of 15€ is added as an incentive. Every participant has to fill in their gender, age, education level, education, math grade and they had to rate their mental arithmetic skills on a 5-point scale from very bad to very good. Education levels are questioned via the Dutch education system. Participants can answer the question with; mbo; hbo or university. Before the official test starts, participants are allowed to see three example questions. After 45 seconds, the first test automatically starts. After the first three and halve minutes, the test automatically stops. A little evaluation follows the first test with questions about how they think they performed in the first test; if they exerted a lot of effort while answering the questions; and about their chances to win the test. Participants can answer these questions on a 5-point scale from very bad to very good.

After this evaluation, every participant receives information about the game they are in. There is a 50% chance of being in the quota group, and a 50% chance of being in the no quota group. Each participant is randomly assigned to one of the groups. When a participant is in the quota competition, he is told that among the 50% winners, at least 30% should be female. As a consequence, if there are not 30% females in the winning group, it means that in order to reach the 30% quota, females from the losing group have to be transferred to the

(13)

winning group. Since the percentage of winners has to remain the same (50%) it means that some men in the winning group will be transferred to the losing group. This could mean that better performing men will be passed by. Therefore, the quota is an advantage for less performing females, and a disadvantage for better performing males. In the no quota group, there is no information provided to the participants about a quota or not. The participants in this group are told to perform the same test again without further information. This group is included as a control group to check if the results found in the first group make sense.

After the second test, there is a second evaluation with more questions. The first three questions are the same as in the first evaluation. Participants are also asked what they think of the gift card as an incentive for this experiment. Extra questions are; what the participant thought about the content of the test; if they would participate again in similar tests; and if they believe their selves to be competitive. These questions can be answered on a 3-point scale varying from No to Average to Yes. All participants are asked; if they thought they performed test two better/worse/not different than test one; and if they prefer male/female/no preference opponents. All participants also have to answer some statements about quotas. The first question is: do you agree if companies make agreements (quota agreements) about how many women should be working in the company? The second statement is: companies who implement a quota have a positive image. If a participant thinks companies who implement quotas have a negative image since the participant probably dislikes a quota, then his incentives to work harder for that specific company may be smaller than if he agrees with quota systems. Adding these two questions gives the option to check if the questions correlate with the score of test 2. For participants in the no quota group, the answers given on these questions should not be correlated with the score of test 2. These questions can be answered on a 5-point scale varying from strongly disagree to strongly agree.

The participants that are in the quota group, are presented two additional questions. First question is: was the quota an incentive for you to exert more/less effort? Second question is: Do you think your probability of winning is higher or lower after the quota is implemented? At last, participants who are interested in the prize are asked for their name and mail address.

(14)

In order to find any results for hypothesis 1, I will do some Wilcoxon rank-sum tests. In this test, it is tested if there are any differences between groups (quota and no quota, males and females). For the second and third hypothesis, I will run OLS-regressions using stata. Below the empirical models used to run the regressions are provided. In section 3.3. the variables used will be further explained.

Model 1: Score test 2 = ∝ + 𝛽1 * Quota + 𝜀

Model 2: Score test 2 = ∝ + 𝛽1 * Quota + 𝛽2 * Ability + 𝛽3 * Female + 𝜀

Model 3: Score test 2 = ∝ + 𝛽1 * Quota + 𝛽2 * Score test 1 + 𝛽3 * Female

+ 𝛽4 * Quota*Female + 𝜀

3.2. Data collection / Research sample

The data used in this paper is collected by running an experiment. In total, there are 57 responses. However, one of the participants did not fill in any answers in the second test. Therefore, this participant is left out of the data set. In Table 1 below, an overview is provided on how many participants per gender, education level and age are in the quota group and in the no quota group. Averages are provided as well.

Quota No Quota Total

quantity average quantity average quantity average

Males 14 0.5 13 0.46 27 0.48 Females 14 0.5 15 0.54 29 0.52 mbo 1 0.04 0 0 1 0.017 hbo 7 1/4 9 0.32 16 0.286 University 20 20/28 19 0.68 39 0.696 Age 19-70 28.5 19-61 29.64 19-70 29.1

Table 1: Quantity and averages of male, female participants in the quota and no quota group. Quantity and averages of participants with an mbo, hbo and university education level per quota group. Interval of age and average age per quota group.

3.3. Data variables

3.3.1. Dependent variable

Score test 1 is used as the measure for individual performance. The test score is calculated by the number of correct answers per individual in test 1.

Score test 2 is also used as the measure for individual performance. The test score is calculated by the number of correct answers per individual in test 2.

(15)

3.3.2. Independent variables

Female is the first independent variable. For this variable a dummy is included. The variable is 1 if the participant is female, and 0 if the participant is male.

Quota is the second independent variable. For this variable a dummy is included as well. The variable is 1 if the participant is in the quota group, and 0 if the participant is in the no quota group.

Quota*Female is the last independent variable. This quota is generated by multiplying Quota and Female. The dummies for Female and Quota still count for this new variable.

3.3.3. Control variables

Ability is the first control variable. Score test 1 indicated the ability each individual has for mathematical problems. Therefore, the variable ability is equal to score test 1. To prevent confusion, the variable is called ability and not score test 1. For hypothesis 1, score test 1 is the dependent variable, but for hypothesis 2 and 3, score test 1 is a control variable. It is used to control for the relation between the ability and the score test 2.

University is the second control variable. This control variable is added to investigate if there is a relation between the education level and the scores of the tests. There are three levels of education; mbo; hbo and University. For this variable a dummy is included. The variable is 1 if the participant has a University education level, and 0 if the participant has a mbo or hbo education level.

Age is the third control variable. This variable is added to control for the relation between age and the test scores. In this experiment, the youngest participants are 19 years old, and the oldest participant is 70 years old.

Math grade is the fourth control variable. Since the test is based on math skills, the math grade is expected to influence the test scores. For this variable, the end grades for mathematics in the last year of high school is used.

Content is the fifth control variable. Participants were asked if they liked the content of the tests. They could answer the question on a 3 point scale, varying from No to Average to Yes. The answers are ranked as 1, 2 and 3 respectively.

(16)

Repeat is the sixth control variable. Participants were asked if they would participate in other similar experiments. They could answer the question on a 3 point scale, varying from No to Average to Yes. The answers are ranked as 1, 2 and 3 respectively.

Quota Agreement is the seventh control variable. Participants were asked if they agree on companies implementing quotas. They could answer the question on a 5 point scale, varying from strongly disagree, to disagree, to neutral, to agree, to strongly agree. The answers are ranked as 1, 2, 3, 4 and 5 respectively.

Image is the last control variable. Participants were asked if they think companies who implement quotas have a positive image. They could answer the question on a 5 point scale, varying from strongly disagree, to disagree, to neutral, to agree, to strongly agree. The answers are ranked as 1, 2, 3, 4 and 5 respectively.

(17)

4. Results

4.1. Descriptive Statistics

In this first section, a clear overview of the data collection is given. Table 1 provides information about the averages of the variables in the quota and no quota group. The table provides a Wilcoxon rank-sum test. The Wilcoxon rank-sum test is a non-parametric test. This test is used for the reason that the number of observations in this experiment is low. The Wilcoxon rank-sum test tests whether there are significant differences between the groups. In other words, it tests whether the randomizer of the experiment worked. If there are no significant differences between the groups, it is easier to conclude if a difference in performance is caused by the quota treatment.

The Wilcoxon rank-sum test, tests if the averages of each variable are the same among both groups. For example the null hypothesis can be: the fraction of females in the quota group is equal to the fraction of females in the no quota group. If the p-value is lower than 5% (0.05), then the null hypothesis can be rejected. On a significance level of 5%, it means that the fraction of female of the quota group is not equal to the fraction of females of the no quota group. Note: this is an example.

The number noted for the variable Female, is the fraction of females in the given group. The number noted for the variable University, is the fraction of participants with a university education level in the given group. The number noted for the variable Score test 1, is the average number of correct answers in the given group. The number noted for Age, is the average age in the given group.

Wilcoxon rank-sum test 1

t-test (p-value)

No Quota Quota No Quota = Quota

Female 0.5357143 0.5 0.791

University 0.6785714 0.7142857 0.7733

Score test 1 7.7 6 0.022

Age 29.64286 28.53571 0.2984

N 28 28

Table 1: Descriptive Statistics. Fraction of females and participants with University education level, the average score test 1 and the average age in the quota and no quota group. P-values of the Wilcoxon rank-sum test.

(18)

The fraction of females is very likely the same in both groups. This is for the reason that the p-value from the test is 0.791. Since this value is greater than 0.05 we cannot reject the null hypothesis. This conclusion is the same for the variables University (p-value = 0.7733) and Age (p-value = 0.2984). This means that there are no significant differences between the groups in the variables Female, University and Age. For the variable Score test 1, the conclusion is different. Since the p-value (0.022) of this variable is lower than 0.05, we have to reject the null hypothesis on a 5% significance level. This means that the ability of the individuals is not the same in both groups.

Second, the differences between men and women are tested as well. To test this, also the Wilcoxon rank-sum test is performed in order to test whether there are significant differences in the variables between men and women. Since this test tests whether there are differences between men and women, the variable Female is excluded. The other three variables in this test, are the same as the three variables in the first test.

The number noted for the variable University, is the fraction of participants with a university education level given the gender. The number noted for the variable Score test 1, is the average number of correct answers of the gender group. The number noted for Age, is the average age of the gender group.

Wilcoxon rank-sum test 2

Mean Mean t-test (p-value)

Male Female Male = Female

University 0.6666667 0.7241379 0.6432

Score test 1 7.3 6.4 0.4319

Age 31.07407 27.24138 0.1783

N 27 29

Table 2: Descriptive Statistics. Fraction of participants with a University education level, the average score test 1 and the average age for males and females. P-values of the Wilcoxon rank-sum test.

Table 1 shows that the variables University (p-value = 0.6432), Score test 1 (p-value = 0.4319) and Age (p-value = 0.1783) are the same for both males and females on a significance level of 5%.

(19)

4.2. Hypothesis 1

Niederle and Vesterlund (2007) state that there will be no large differences in performance between men and women. Therefore, this statement is included in this paper as the first hypothesis. The performance is measured with the results of Score test 1. Only the results of test 1 can be used to measure any difference between groups, since the information set in test 2 is not the same for all participants. This is due to the quota treatment or not. In the previous section, the Wilcoxon rank-sum test is performed. This test tests whether there are differences on the level of university education level, score test 1 and age between males and females. In Table 2 of section 4.1. you find the descriptive statistics of this test.

The null hypothesis of this test is: The will be no large differences in performance between men and women. So the average score of men is equal to the average score of women. The alternative hypothesis states that there will be large differences in performance between men and women. The p-value found in the Wilcoxon rank-sum test is 0.4319. Since this p-value is greater than 0.05, we cannot reject the null hypothesis. Therefore, with a significance level of 5%, we can conclude that there are no large differences in performance between men and women.

4.3. Hypothesis 2

The second hypothesis of this paper is: A quota treatment has an effect on participant’s performance. It will be tested if the quota has an effect on the score of test 2 or not. The first model used for hypothesis 2 is:

Score test 2 = ∝ + 𝛽1 * Quota + 𝜀

The dependent variable is Score test 2, and the independent variable is the quota. A constant and an error term are added to the model. In this model, we test if the quota treatment has an effect on the performance or not. The null hypothesis and the alternative hypothesis are given below:

H0: 𝛽1 = 0

H1: 𝛽1≠ 0

(20)

coefficient of the quota is equal to zero. If the p-value of the t-test is lower than the significance level of 5% (0.05), than we can reject the null hypothesis. That means that the coefficient is significantly different from zero and therefore has an effect. In Table 3 you will find the regression analysis. For this model we only have to look at the first regression.

Dependent variable: performance test 2

(1) (2) (3) Quota -0.8928571 0.008188 -1.074247 (0.8304656) (0.6969693) (0.9666681) Ability 0.5623273*** 0.5661569*** (0.1129532) (0.1113503) Female -1.200119* -2.250806** (0.6773097) (0.9388641) Quota*Female 2.102678 (1.321201) Constant 7.107143 3.432193 3.965656 (0.5872278) (1.095166) (1.130223) Summary Statistics F 1.16 10.98 9.11 Prob > F 0.2871 0.000 0.000 R2 _0.021 _0.3877 _0.4167 R2_adj _0.0028 _0.3524 _0.3709 N 56 56 56 *** P < 0.01 ** P < 0.05 * P < 0.10

Table 3: Regression analysis of the relationships between Quota, Ability and Female on the Score of test 2.

Table 3 shows three regressions. The first regression is denoted as (1). The stars notated after the coefficients indicate whether the p-value is smaller than the significance level. According to this regression, the participants in the no quota group obtain a score in test 2 of 7.11 answers, while the participants in the quota group obtain a score of 7.11 – 0.89 = 6.22 answers. The p-value of the t-test of the Quota coefficient is 0.287, thus larger than 0.05. We cannot reject the null hypothesis. Therefore, we cannot say whether the quota treatment has an effect on the performance.

For this reason, we add control variables to the regression. The second model used to test hypothesis 2 is:

(21)

The dependent variable is Score test 2, and the independent variables are Quota, Ability and Female. A constant and an error term are added to the model. In this model, we test if the quota treatment has an effect on the performance or not, controlled for ability and gender. The null hypothesis and the alternative hypothesis are given below:

H0: 𝛽1 = 0

H1: 𝛽1≠ 0

The null hypothesis in this model remains the same as the hypothesis in the previous model. We still want to test whether the coefficient for Quota is equal to zero or not, since we want to find out if the Quota has a significant effect or not.

The second regression is denoted as (2) in the table. As Table 3 shows, the effect of Ability is significant on a 1% significance level, and the effect of Female is significant on a 10% significance level. It is very interesting that the variable female has a significant negative effect on the score of test 2, since hypothesis 1 showed that the performance for men and women are the same for test score 1. The p-value of the Quota coefficient is 0.991. Therefore, we cannot reject the null hypothesis, and cannot say whether the Quota treatment has an effect on performance.

The conclusion of hypothesis 2 is that there is no significant effect of quota on performance for this data set. Even when controlling for the variables Ability and Female, no significant effect was found.

4.4. Hypothesis 3

The third hypothesis of this paper is: Does the quota have a different effect for male and female participants? To test this hypothesis, we use the same model as in hypothesis 2, only another variable is added. This variable is Quota * Female. The model used for hypothesis 3 is:

Score test 2 = ∝ + 𝛽1 * Quota + 𝛽2 * Score test 1 + 𝛽3 * Female + 𝛽4 * Quota*Female + 𝜀

The dependent variable is Score test 2, and the independent variables are Quota, Ability, Female and Quota*Female. A constant and an error term are added to the model. The effect the quota has on male participants will be 𝛽1. The effect the quota has on female

(22)

participants will be 𝛽1 + 𝛽4. 𝛽4 shows the difference in performance between men and

women. The null hypothesis and the alternative hypothesis are given below:

H0: 𝛽4 = 0

H1: 𝛽4≠ 0

In Table 3, the outputs of the third regression (3) are provided. The table shows that the coefficients of Ability and Female have significant effects. Ability is still significant on a 1% significance level, and the effect of Female is now significant on a 5% significance level. In this regression, the variable female again has a significant negative effect, which is strange, since the performance of men and women was tested to be the same. The p-value of the Quota coefficient is 0.272 and the p-value of the Quota*Female is 0.118. Thus both coefficient do not have a significant effect on the score of test 2.

In Appendix B, you will find a regression analysis of the effect of some of the survey questions on test score 2. Since the effects are not significant at all, the relations between the variables and score test 2 is not relevant for this thesis. Therefore, it is not discussed in this section.

(23)

5. Discussion

5.1. Findings

The aim of this paper is to answer the question whether a quota system has an effect on individual performance or not. To find an answer to this question, three hypothesis were tested. The results from the experiment do not show significant effects of a quota.

The first hypothesis predicted that there are no differences in performance between men and women. The observable features per gender were compared to each other to test whether there were differences between men and women. The results show that there are no statistical significant differences between the performance of men and women. Therefore, we can accept the hypothesis. This is in line with the findings of Niederle and Vesterlund (2007). However, it is important to note that there are differences in ability between participants in the quota and in the no quota group. Results show that the participants in the no quota group are significantly better in this mathematical test than participants in the no quota group.

The second hypothesis predicted that a quota system has an effect on performance. According to the first regression, where only the effect of a quota on performance was tested, the coefficient of quota was negative. However, this coefficient is not statistically significant. The second regression tested the effect of quota on performance, controlled for the ability and gender. In this regression, the quota has a very small positive effect on the performance. However, this effect is not statistically significant either. In contrary to hypothesis 1, the regressions show that females perform significantly less than males do in test 2. In the third regression, the effect of a quota on performance was tested, but controlled for the ability, the gender and the interaction term of quota times female. To investigate whether men and women react differently to a quota treatment, this regression was added. The effect of the quota was negative, though this is not statistically significant. The interaction term of quota times female can say something about the difference in performance between men and women. The results show that female participants in the quota group scored higher in test 2 than participants that are not female and/or in the no quota group. Nevertheless, these findings are not statistically significant either.

(24)

Since most of the effects are not statistically significant, we cannot conclude whether the quota system has a positive or negative effect on individual performance. Nor can we say if a quota system has an effect at all.

5.2. Contributions and implications for the literature

Until now, most studies did research on how a quota system affected the entry decisions in tournaments (Niederle et al., 2013: Balafoutas and Sutter, 2012). The study in this paper is different, since it investigates whether individuals perform different when they know they compete in an environment where a quota is implied in order for at least 30% women to be among the winners. Although this paper did not find any significant effects, it may be helpful to know whether a quota has an effect on performance or not.

5.3. Practical implications

If the results found were significant, than we would have been able to say more about the effect a quota system has or not. If the quota indeed has an effect on performance it can be relevant for companies who implemented a quota, or are considering a quota to select new employees. If the performance of men and women differ under a quota system, companies can take this effect into account when deciding about who to employ. This is beneficial for the company for the reason that with this knowledge they can select the best employees.

5.4. Limitations

This research had some limitations. To begin with the number of observations. The number of observations in this experiment was very low, so it is harder to find significant effects. The experiment was in Dutch, so to reach more participants, it would have been better if the experiment was in English.

Second, the incentive for the participants was in my opinion not high enough. Participants will be more motivated to work hard when the incentive is high. The probability of winning may be too low as well, since only 1 participant of the whole game could win the tournament.

(25)

A third limitation is that in the introduction of the experiment, participants were told that they were taking part in a hypothetical job interview. It would have been better if they were told to just participate in a game, without hypothetical information.

Fourth, in the second round of the experiment, the participants in the quota group were provided with information about the quota and how this influences the probabilities of winning the game. On the other hand, the no quota group was not provided with any information. It would have been better if the participants were reminded about the winning settlement.

The last limitation is that some of the extra question were asked on a 3 point scale. The answers would have been more detailed when the questions were asked on a 5 or 7 point scale.

5.5. Directions for future research

For future research, I recommend to do a similar experiment, but adjusting for the limitations discussed before. First of all, the sample should be larger. Second, participants should be able to win a higher prize when they win the tournament and the probability of winning should be higher. Third, the introduction should only tell that participants are part of a tournament, and the probability of winning the tournament should be provided. Fourth, participants in the no quota group should have been reminded about the winning probability before the second test starts. And last, the additional survey should be more detailed.

(26)

6. Conclusion

The main goal of this paper was to answer the question whether a quota has an effect, either positive or negative, on individual performance or not. This was divided into three hypothesis. The first hypothesis tests whether there is a difference in performance between men and women in this experiment. Results show that the scores of men and women is significantly the same. Interesting is that the ability of the participants in the no quota group is significantly higher than the ability of the participants in the quota group. The second hypothesis tests whether the quota has an effect on performance. The results show both positive and negative effects, but in both regressions the effects are not significant. Therefore, we cannot conclude anything about the effect of a quota system on performance in this study. The third hypothesis tests whether men and women perform different under a quota system. Results show that females in the quota group perform better than other participants, but this result is not significant.

(27)

References

Agnew, J. R., Anderson, L. R., Gerlach, J. R., and Szykman, L. R., 2008. Who Chooses Annuities? An Experimental Investigation of the Role of Gender, Framing, and Defaults. American Economic Review. Vol. 98: 418 – 442.

Akyol, M. et al., 2015. A tradable employment quota. Labour Economics. Vol. 36: 48-63.

Balafoutas, L., and Sutter, M., 2012. Affirmative Action Policies Promote Women and Do Not Harm Efficiency in the Laboratory. Science. Vol. 335 (6068): 579 – 582.

Brett, J. F., Heilman, M. E., and Rivero, C. J., 1991. Skirting the Competence Issue: Effects of Sex-Based Preferential Selection on Task Choices of Women and Men. Journal of Applied Psychology. Vol. 33: 99 -105.

Charness, G., and Gneezy, U., 2011. Strong Evidence for Gender Differences in Risk Taking. Journal of Economic Behavior & Organization. Vol. 83(1): 50 – 58.

Croson, R., Gneezy, U., 2009. Gender difference in preferences. Journal of Economic Literature. Vol. 47 (2): 448–474.

Dorrough, A. et al., 2016. Revealing side effects of quota rules on group cooperation. Journal of Economic Psychology. Vol. 57: 136-152.

Eckel, Catherine C., Grossman, Philip J., 2008. Men, women and risk aversion: experimental evidence. In: Plott, C., Smith, V. (Eds.), Handbook of Experimental Economics

Results., vol. 1. Elsevier, New York, pp. 1061–1073, Chapter 113.

European Commission, 2012. Impact assessment on costs and benefits of improving the gender balance in the boards of companies listed on stock exchanges. Brussels:

(28)

European Commission. Retrieved from <http://ec.europa.eu/justice/gender -equality/files/womenonboards/impact_assesment_quotas_en. pdf>.

Hartog, J., Ferrer-i-Carbonell, A., and Jonker, N., 2002. Linking Measured Risk Aversion to Individual Characteristics. Kyklos. Vol. 55: 3 – 26. 

Heilman, M.E., Block, C.J., Lucas, J.A., 1992. Presumed incompetent? Stigmatization and affirmative action efforts. Journal of Applied Psychology. Vol. 77: 536.

Heilman, M.E., Block, C.J., Stathatos, P., 1997. The affirmative action stigma of

incompetence: effects of performance information ambiguity. Academic Management Journal. Vol. 40: 603–625.

Heilman, M.E. and Mardenfeld Herlihy, J., 1984. Affirmative Action, Negative Reaction? Some moderating Conditions. Organizational Behaviour and Human Performance. Vol. 33: 204-213.

Hermalin, B. and Weisbach, M., 2003. Boards of directors as an endogenously determined institution: a survey of the economic literature, Federal Bank of New-York Economic Policy Review. Vol. 9: 7–26.

International Labour Organization, 2012. Global employment trends 2012. Geneva: International Labour Organization. Retrieved from <http://www.ilo.

org/global/research/global-reports/global-employment-trends/WCMS_171571/lang -en/index.htm>. 

Niederle, M. and Versterlund, L., 2007. Do women shy away from competition? Do men compete too much? The Quarterly Journal of Economics. Vol. 122 (3): 1067-1101.

(29)

Niederle, M., Segal, C., and Versterlund, L., 2013. How costly is diversity? Affirmative action in light of gender differences in competitiveness. Management Science. Vol. 59 (1): 1 -16.

Powell, M., and Ansic, D., 1997. Gender differences in risk behaviour in financial decision-making: An experimental analysis. Journal of Economic Psychology. Vol. 118: 605 – 628.

Ross-Smith, A. and Bridge, J., 2008. ‘Glacial at best’: women’s progress on corporate boards in Australia, in Women on Corporate Boards of Directors, Vinnicombe, S., Singh, V., Burke, R. J. and Huse, M. (Eds), Edward Elgar, Northampton, MA, p. 6.

Sabatier, M., 2015. A women’s boom in the boardroom: effects on performance? Applied Economics. Vol. 47: 2717 – 2727.

Sealy, R., Singh, V. and Vinnicombe, S., 2007. The Female FTSE Report 2007, Cranfield, UK.

Singh, V., Terjesen, S. and Vinnicombe, S., 2008. Newly appointed directors in the

boardroom: how do women and men differ? European Management Journal. Vol. 26: 48–58. doi:10.1016/j.emj.2007.10.002

Van Vugt, M., De Cremer, D., and Janssen, D. P., 2006. Gender Differences in Cooperation and Competition. Psychological Science. Vol. 18 (1): 19 – 23.

(30)

Appendix A

Correlation

Score test 1 Female University Age grade Math

Score test 1 1.0000

Female -0.1444 1.0000

University 0.2342 0.0515 1.0000

Age 0.2238 -0.2038 -0.1412 1.0000

Math grade -0.0293 0.1627 0.1352 -0.0831 1.0000

Correlation between score test 1, female, university, age and math grade.

Appendix B

Dependent variable: score test 2

(1) (2) Content 1.149927 0.5153528 (0.6004877) (0.5114899) Repeat -0.1218994 -0.5346689 (0.6361698) (0.5446996) Quota Agreement 0.4019378 0.7476099 (0.5539764) (0.4799099) Image -0.7111258 -0.8052052 (0.6818865) (0.6047926) Quota 0.3548459 (0.7563682) Ability 0.6184704*** (0.1299862) Female -0.9121997 (0.7445395) Constant 5.230039 3.068934 (1.710024) (1.58192) Summary Statistics F 1.35 Prob > F 0.2632 R2 0.0959 R2_adj _0.025 N 56 56 *** P < 0.01 ** P < 0.05 * P < 0.10

Regression analysis on the effects of content, repeat, quota agreement, image, quota, ability and female on test score 2.

The effect of a quota on individual performance