The lucky one : giving behaviour in a real effort experiment

(1)

The Lucky One; Giving

Behaviour in a Real Effort

Experiment

Noortje Boogers 10248021 5 August 2016 Abstract

This paper investigates how a positive external shock influences giving behaviour in a real effort experiment. The results show that when subjects received the positive external shock they tend to show more generous behaviour. Furthermore, it is found that when a positive external shock is present but not received by subjects, giving behaviour does not seem to differ from the case were only effort determined the final earnings. The results are consistent with predictions derived from literature. These findings are relevant for both governments and charity funds to promote generous behaviour.

Prof. Dr. J.H. Sonnemans 2015/2016

Master Thesis Seminar Semester 2, period 3 Behavioural Economics

(2)

2

STATEMENT OF ORIGINALITY

This document is written by Noortje Boogers (10248021) who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

1. INTRODUCTION

A recent study by Côté et al. (2015) showed evidence that high income individuals tend to be less generous in highly unequal economic areas than in less unequal economic areas. They explained this observed behaviour by stating that it could be the case that high income individuals perceived the current inequality as just, because they worked harder than the lower income individuals. This could raise some questions about fairness. Under what circumstances is it fair that one person earns more than one other, just because he is higher educated, for example? And could one person’s salary level be perceived as a reward for pure effort or as a reward for a combination of both effort and luck? The world’s wealthiest individuals and families are already aware of how luck influenced their final income. For example, Warren Buffet states the following about this:

“My wealth has come from a combination of living in America, some lucky genes, and compound interest. Both my children and I won what I call the ovarian lottery. (For starters, the odds against my 1930 birth taking place in the U.S. were at least 30 to 1. My being male and white also removed huge obstacles that a majority of Americans then faced.)”

It thus can be stated that the awareness of luck in determining final income plays a role for the wealthiest individuals to give the majority of their wealth to philanthropy1. This could make one wonder whether the awareness of luck could influence giving behaviour for other people as well.

A lot of laboratory and field experiments have shown that people are willing to help others by making personal costs (Adreoni, 1989; Krebs, 1970). Despite that, none of the research so far investigated the relationship between how people obtained their income – via pure effort or via effort and luck – and giving behaviour.

This study investigates this by answering the following research question: “Does giving behaviour differ when luck influences final earnings compared to when earnings are determined by effort only?”. In other words, when a positive external shock is salient, do people tend to show more generous behaviour compared to a situation when a positive external shock is not present?

1_{On the website givingpledge.org the world’s wealthiest individuals and families declare that they have made}

(5)

5

To answer this research question, an online experiment has been conducted. This research finds that subjects tend to be more generous when they receive a positive external shock. This result could be interpreted in two ways. First, it could be the case that people tend to care more about fairness when they are positively affected by luck. Second, it could be that fairness preferences change when luck is made salient, in the sense that fairness preferences are context dependent. Furthermore, when a positive external effect is present but not received, giving behaviour does not seem to differ from cases where income is earned by effort only.

Answering the research question is not only important in terms of academic achievement, the results of this study could also help both governments and charity funds. It could advice governments how to promote redistribution, and it could help charity funds to recruit more financial donors.

The structure of this study is as follows; the next chapter describes related literature and its main findings. In the third chapter the methodology is discussed. This is followed by a chapter in which the results are analysed. After that these results are discussed and a few shortcomings are described. The final chapter of this study provides a short summary of the main findings and ends with some concluding remarks.

2. LITERATURE REVIEW

This chapter gives an overview of the research about the relationship between generosity and a combination of both effort and luck. Because generosity is hard to measure, giving behaviour in a dictator game is used as a proxy for generosity. First, a short review is given about related literature describing giving and altruistic behaviour in general. Secondly, research about the relationship between giving and generated income is described. After that, the literature which investigates how external shocks affect giving behaviour in real effort experiments is discussed. This is followed by a section which gives a short overview of different fairness preferences. The final section describes literature about the locus of control and beliefs about luck.

A. GIVING AND ALTRUISTIC BEHAVIOUR

Instead of acting only in one’s own interest, altruistic people tend to be selfless caring individuals, who derive utility from sacrificing something to give to someone else (Erkal et

(6)

6

al., 2009). Because standard economics predict all individuals to be perfectly selfish, this kind of behaviour is a highly studied subject by experimental economists.

In the research called “Why do people give?” Versterlund (2001) investigates the various reasons behind altruistic behaviour. Based on theoretical and experimental research she states that there are two reasons why people show altruistic behaviour.

The first reason is public in nature, which means that both the person who gives (from now this is called the benefactor) and other individuals benefit from a donation. The benefactor may, for example, care about homeless people. By giving money to a fund which takes care of the homeless people, the benefactor benefits, and the homeless people benefit as well. Although the derived utility from the output of charity funds is a convincing explanation for altruistic behaviour, this is not the only reason why people tend to give (Vesterlund, 2001). People who do not contribute to a charity fund are also benefited by the donation of one individual, therefore giving to charity can also be seen as a public good. When the benefits for a donator are the same as the benefits for a someone who does not contribute, there is a strong incentive to free-ride (Samuelson, 1954; in Vesterlund, 2001). This incentive convinced researchers that there must be another reason why people show altruistic behaviour.

The second motive to show altruistic behaviour is private in nature. Andreoni (1989) argues that people not only tend to give because of pure altruistic preferences, but also because they get a “warm glow” from giving. So instead of giving to help others, people might be giving to get a good feeling about themselves. If this is the case, they no longer view the donations by others as perfect substitutes for their private donations. This cancels out the free-rider incentive discussed before (Vesterlund, 2001). Besides the “warm glow”, literature proposed several other private benefits that people might experience from giving, among which; signalling wealth in a socially acceptable way (Glazer & Konrad, 1996; in Versterlund, 2001), “buying” the goods they receive from charity funds or assuring that they can feel good about themselves because they have given their just share (Rose-Ackerman, 1982; in Versterlund, 2001).

In economic experiments, one way to test altruism is by the dictator game. This game has two players, a dictator and a receiver. The game starts with a pie which has to be divided between the dictator and the recipient. The dictator than dictates the division of the pie and the receiver just receives his share. Standard game theory predicts that the dictator keeps the whole pie to himself, because the recipient has no opportunity to decline the offer (Camerer,

(7)

7

2003). Despite these predictions, it is often found that in experiments with full anonymity still 60% of the participants tend to give part of their wealth to others (Hoffman et al, 1994). According to Carpenter et al. (2006) this could be due to the fact that subjects come to the experimental lab to play a game, and not to take it all themselves. As a way to avoid this kind of behaviour they suggest to let subjects earn their income by a real effort task. This is discussed in the next section.

When analysing generosity in economic and social experiments different giving behaviour is observed. People show different kinds of altruistic behaviour because of various reasons. The reasons why giving behaviour differs can be categorized in four different levels of generality; temporary psychological states, personal traits, social roles and demographic variables, and social norms (Krebs, 1970).

These levels explain giving behaviour in different ways. Temporary psychological state effects tend to tell little about the benefactor. Furthermore, personal traits tend to give a better insight. When observing giving behaviour social roles and demographic variables tend to show even more general, permanents and basically characteristics (Krebs, 1970).

These differences are mostly observed in economic experiments. For example, Adreoni and Vesterlund (2001) performed an experiment to test whether gender affects altruistic behaviour and fairness principles. They find that women show more altruistic behaviour when the price of giving is high, while men give more when the price of giving is low. When the price of giving is one on one (when you give one euro to someone, the other person can spend one euro), there seem to be no differences between men and women.

The current study mainly focuses on how temporarily behavioural states can influence altruistic behaviour. This means that it is mainly investigated how giving behaviour is affected when final earnings are influenced by a positive external shock.

B. EFFECTS OF INCOME

Including income in the analysis of the dictator game might explain why experimental findings differ from game theoretical predictions. In the dictator game the subgame equilibrium is to offer nothing (Camerer, 2003). Nevertheless, in experiments it is often found that 60% of the subjects give away part of their wealth (Hoffman et al, 1994).

When subjects have to exert effort to earn their income, they tend to show more selfish behaviour. In their experimental research, Cherry et al. (2002) find that when people had to work for their money, other-regarding behaviour is reduced. This result was mainly

(8)

8

found when subjects could earn high rewards as a result from hard work. This suggests that when people have to exert effort to earn their money, they show more selfish preferences and game theoretic predictions might be more accurate (Oxoby & Spraggon, 2008). When the dictator game is extended to a n-player game, a self-selection mechanism – in the sense that the most selfish people work the hardest – could also explain the increase in selfish behaviour (Erkal et al., 2011).

When subjects have to compete to earn the right to be dictator, they tend to show even more self-regarding behaviour. This shows that, besides exerting effort to ‘earn the pie’, also entitlements play an important role in determining giving behaviour (Hoffman et al., 1994). However, different results are found when only the recipient has to exert effort to determine the size of the pie. Experimental evidence shows that in this case dictators are less selfish, and on average give the exact amount the recipient ‘added’ to the size of the pie (Oxoby & Spraggon, 2008). Nevertheless, when exerted effort by the recipient is perceived as too low by the dictator, recipients tend to be “punished” (Ruffle, 1996). It thus can be stated that when subjects have to work for the size of the pie they do not express selfish behaviour only because they worked hard for it, they also tend to do this because they feel they earned the property and entitlement rights.

C. EFFECTS OF EARNED INCOME AND AN EXTERNAL SHOCK

In real life income is not determined by effort only, besides effort external factors seem to influence one’s earnings as well (Tonin & Vlassopoulos, 2016). It therefore could be useful to look at how a combination of both effort and external shocks influences giving behaviour.

The evidence of the effects of a negative external shock for the recipient on giving is mixed. In an experimental study Fong (2007) finds, that when welfare recipients are poor because of bad luck instead of a lack of effort, they tend to receive more from dictators. This is supported by other experimental findings (Erkal et al., 2011). However, Sheremeta et al (2011) find that when comparing giving behaviour in a two-person dictator game, subjects who exerted high effort tend to give more to subjects who exerted low effort instead of subjects who had bad luck.

When the benefactor receives a positive external shock there is no clear consensus about the findings either. In their experimental study, Erkal et al. (2011) find that when subjects are positively affected by luck their giving behaviour does not differ from when they are not affected by luck. These findings are confirmed by a study of Tonin and Vlassopoulos

(9)

9

(2016). They find that when subjects are positively affected by luck – in a way that they receive a higher piece rate than other subjects – they tend to give the same absolute amount as subjects who received a lower piece rate. Nevertheless, other research shows that when subjects receive a positive external shock, they tend to give more (Sheremeta et al, 2011). A possible explanation for this mixed evidence could be the different kind of external shocks used in different experiments. Becker (2013) investigated the relation between redistribution behaviour in case of different kinds of luck. As different kinds of luck they used wage luck (subject get different wages for extended effort), endowment luck (subjects get an effort-unrelated fixed amount) and payment luck (subjects receive probabilistic payments). It is found the effects from different kinds of luck seem to differ. High performers tend to take more for themselves in case of wage luck and low performers tend to take more for themselves in case of endowment luck (Becker, 2013).

D. FAIRNESS PREFERENCES

Many dictators in a dictator game seem to have fairness considerations. Based on theoretical models and different giving patterns subjects show in experiments, several types of subjects can be classified.

The first type is the self-regarding utilitarian. This is a person who acts according to the predictions from standard economic theory; someone who maximizes his own earnings (Hoffman & Spitzer, 1985). This person’s utility is independent of any other person’s utility and regards every situation in which he gets more money as more desirable.

The second type classified is the egalitarian. Although standard economic theory predicts all people to be utilitarians, Fehr and Schmidt (1999) show that contrary to these predictions not all people seem to behave this way. It is possible to make a distinction between two types of egalitarians (Hoffman & Spitzer, 1985); people who care about equality and people who care about equity (Konow et al., 2009). The former tends to care about whether final outcomes are equal. The latter seems to care about the accountability principle; fair outcomes should be related to the contributions which are in one’s control, like effort (Konow, 2000). In anonymous experiments subjects tend to care mostly about equity instead of equality (Krawczyk, 2010), but when the anonymity conditions are lifted this fairness ideals tend to move progressively from equity towards equality (Konow et al, 2009).

(10)

10

E. BELIEFS ABOUT LUCK

Because the current study mostly investigates the effects of luck on giving in a real effort experiment, it is important to take beliefs about luck in to consideration as well. Only a few researchers looked at the beliefs about luck and especially how beliefs about luck influences giving behaviour.

Sheremeta et al. (2011) looked at what kind of beliefs subjects had about other’s luck. In a real effort experiment where final earnings were influenced by luck, they compared American and Spanish subjects. When subjects did not know how luck influenced other’s earnings they found that when Spanish subjects observed low earnings they typically believed that someone had suffered from bad luck, while Americans had the beliefs that this person did not work hard enough.

In an attempt to measure how beliefs about one’s own luck in general influences giving behaviour, this study makes use of the test of ‘Locus of Control’. The locus of control questionnaire was designed by Rotter (1966) to test to which extent people perceive rewards to be an achievement of their own performance. The locus of control makes a distinction between two extremes; beliefs in external control and beliefs in internal control. Beliefs in external control means that subjects believe that achievements obtained are only a result of chance, luck, fate, or because they are surrounded by a great complexity of powers. Internal control beliefs mean that subjects believe that all the rewards and events that happen are because of their doing and the choices they made. Typically, subjects’ beliefs do not fall in one of these extremes, but tend to vary in degree (Rotter, 1966).

When linking the locus of control to economic behaviour some interesting findings are revealed. Furnham (1986) found that female subjects tend to have a more external locus of control than male subjects. Furthermore, it was found that in relation to academic achievement people with internal control beliefs typically tend to obtain higher academic achievements than people with external control beliefs (Findley & Cooper, 1983). In addition, U-shaped evidence for the relation between the locus of control and income was found (Furnham, 1986). This means that both the people with the lowest income as the people with the highest income tend to have an external locus of control.

The validity of the locus of control test is being questioned by several researchers (Furnham, 1986; Kestenbaum and Hammersla, 1976). For example, it is possible that beliefs according to the locus of control are not the cause of someone’s position in terms of income, but the consequence (Furnham, 1986). Besides that, Kestenbaum and Hammersla (1976)

(11)

11

argue that in a forced choice paradigm, a distinction should be made between two equally social desirable options. This is something that Rotter’s locus of control questionnaire fails to do. Although Rotter’s locus of control is criticized, the questionnaire still seems to be highly used in social science.

3. METHODOLOGY

This chapter describes two related experiments and the differences with the current study. Then, it provides a detailed description of the experiment performed for the current study, including a motivation for the choices made. The chapter concludes with a formulation of the hypotheses.

A. RELATED EXPERIMENTS

This study uses a combination of the experiments conducted in the paper of Erkal et al. (2011) and in the paper of Sheremeta et al. (2011).

In the study of Erkal et al. (2011) they try to link relative earnings to giving behaviour in a real effort experiment. By letting four subjects participate in a competition – in the form of a real effort experiment – subjects determined their earnings. The person who earned the most points received the highest amount, the person with the second most points received the second highest amount and so on. Just as in the current study, they had a luck treatment, which affected the height of the score. In this treatment, some subjects received a randomly assigned negative shock, which reduced their points by 30%.

The experiment in the current study is similar in the following aspects. First, it also uses a competition structure to determine the final earnings. Second, in both experiments the shock is randomly assigned and not based on performance.

There are some differences between the experiment of Erkal et al. (2011) and the experiment conducted in the current study as well. First, in the current study subjects have been receiving a positive luck effect instead of a negative luck effect. This is because the current study looks into how the saliency of luck influences giving behaviour in a real effort experiment. Second, in terms of experimental design, subjects did not know beforehand that they had to make a decision about giving away their earnings. This is something they did know when participating in the experiment of Erkal et al. (2011), which resulted in a self-selection mechanism for the most selfish subjects. The third difference with the study of Erkal et al. (2011) is that in the current study subjects only had to compete with one instead

(12)

12

of three other subjects. The final difference is that in the current study subjects had to earn entitlement rights. In the study of Erkal et al. (2011) one transfer decision was randomly implemented, which may have caused some free rider incentives.

The second paper which is closely related to the current study, is the study of Sheremeta et al. (2011). When looking at social sharing norms they made a comparison between US and Europe. First subjects had to participate in a real effort task, in which the payment procedure was as follows: for each correctly solved puzzle they earned a token. After participating in the real effort task a random number of tokens was added to the final score, which could take the values ranging from -50 to +50. In the second stage of the experiment subjects were told that they had the possibility to redistribute their earnings in a dictator game. For each pair of subjects one decision was randomly chosen by the computer to be implemented (Sheremeta et al., 2011).

The current study differs from the study of Sheremeta et al. (2011) in the following aspects. First, instead of randomly selecting one decision, in the current study subjects had to earn the right to be the dictator. Furthermore, the current study focuses on the effects of a positive external shock instead of a combination of both a negative and a positive external shock. By making the positive external shock more salient and assuring subjects that at least one of them will receive this bonus, allows for control of the positive effect only. Finally, the current study makes use of a control group to check whether the existence of an external shock changed giving behaviour when subjects did not receive the bonus anyway.

B. EXPERIMENTAL DESIGN

The experiment of the current study has been conducted online. The experiment has been conducted online to get a larger and more diverse subject pool, which is easier to obtain online than in the lab.

The experiment consisted of two treatments, a control treatment and a bonus treatment. Both treatments used a two-stage game. In the first stage subjects had to participate in a real effort task and in the second stage subjects had to make a decision about what to give. It was decided to include a control group to check whether giving behaviour in the treatment group would not differ when the bonus was present but not received.

Before starting the experiment subjects were told that they were participating in an experiment about decision making. Furthermore, they were told that they would be matched with one other participant and that they had the chance to earn some money based on the

(13)

13

decisions of themselves and the partner they were matched with. In addition, subjects were told that one pair of participants would be selected for payment. To make sure people would not behave in a way that is desirable for the experimenter, subjects were told that the experiment was fully anonymous (Hoffman et al, 1996). To ensure this subjects were told that the supervisor of this thesis, and not the experimenter, would select one pair for payment.

The experimental design is discussed as follows. First, the first stage of the experiment is described because this is the same for both groups. Second, the second stage of the control group is discussed, which is followed by a description of the second stage of the treatment group. After that the questionnaire is discussed. Figure 2 provides a summary of both treatments. For the exact instructions of the experiment, see the appendix.

I. FIRST STAGE

In the first stage of the experiment subjects had to participate in a real effort task in which they had to solve some ‘puzzles’. These puzzles existed of five by five matrices which were filled with randomly generated two digit numbers (Figure 1). Subjects were told that they had to find the highest number in each matrix and add them up. Furthermore, they were told that they had only one attempt to provide the correct answer. For calculating these summations subjects could use a calculator. Before starting with the real effort task subjects had to answer three control questions, to make sure they understood everything. After finishing these questions, subjects got a reminder to get their calculator. Each subject had three minutes to solve as many puzzles as possible and faced a maximum of 30 puzzles. On top of the screen they could see how many puzzles they solved right and wrong, and on the right they could see how much time was left. This task was used before in an experiment of Weber and Schram (2016).

Subjects were encouraged to work as hard as possible by letting them compete for the entitlement rights. They were told that for solving each matrix correct they could earn a point and that the person who earned the most points would be selected for payment. A tournament structure was chosen because of the following reasons. First, most of the previous studies which used a real effort task to determine income also had a tournament design (e.g., Erkal et al, 2011; Hoffman et al. 1994). Also using a tournament structure makes it easier to relate the findings of this study to previous findings. Second, using a tournament structure instead of a piece-rate scheme, allows to make luck more salient, which in turn diminishes the self-serving bias found in the study of Tonin and Vlassopoulos (2016).

(14)

14

Furthermore, to avoid the self-selection mechanism found in Erkal et al. (2011), subjects were not told they had the opportunity to give part of their earned money away in the second stage of this experiment.

II. SECOND STAGE – CONTROL GROUP

In the second stage of the control treatment subjects were asked if, in the case they earned the most points, they would be willing to give something to the other person. Because the participants did not participate in the real effort task at the same time, they did not get any information about the performance of the other subject. It was made clear that the decision they were going to make would only be implemented in case the earned the most points. Subjects could choose from a list of options ranging from “You get €10 and the other person gets €0” to “You get €0 and the other person gets €10”. After this subjects were asked to fill in a short questionnaire, which is discussed later.

(15)

15

III. SECOND STAGE – TREATMENT GROUP

The second stage of the treatment group was slightly different. After finishing the real effort task, subjects were told how many puzzles they solved correctly and that a random noise of three points would be added to the final score. It was made explicitly clear that always one participant would be selected for the bonus, and that the chance to receive the bonus did not depend on the points earned in the real effort task. To make sure subjects understood the bonus, some additional control questions were asked.

After answering the control questions the subjects were asked if, in the case they had the highest final score, they would be willing to give something to the other person. Just as in the control group, subjects did not get any information about the participant they were matched with. In addition, they did not get any information about whether they received the three extra points. Therefore, subjects were told they had to make two decisions, in case they had the highest final score without receiving the bonus and one with receiving the bonus.

Again subjects could choose from a list of options and before picking one option, it was stated one more time that their decision would only be implemented in the case they had the highest final score. After making these two decisions subjects were asked to fill in a short questionnaire.

IV. QUESTIONNAIRE

Both treatments ended with a short questionnaire about some basic characteristics such as gender, age, highest level of education and monthly income. Furthermore, it was asked if subjects participated in an economic experiment before, as this tends to influence behaviour in economic games (Hoffman et al, 1994; Carpenter et al, 2006).

To check whether subjects exerted effort in the real effort task they were asked how hard they worked on a scale from 1 to 5 (where 1 is not hard at all and 5 is really hard), and how they perceived the level of the task, ranging from very easy to very difficult. In both treatments subjects were asked how they decided what to give to the other participant. In the bonus treatment subjects were asked this question two times (one in case they did not receive the bonus, one in case they did receive the bonus). Furthermore, subjects in the bonus treatment were asked how they felt after finding out about the random bonus, as this emotional state might have influenced their behaviour (Krebs, 1970).

(16)

16

The questionnaire concluded with a locus of control survey, based on a design by Rotter (1966). In this survey subjects had to choose between two statements for thirteen times. As an example, one of the statements where subjects had to choose from was:

A. Becoming successful is a matter of hard work; luck has little or nothing to do with it B. Getting a good job depends mainly on being in the right place at the right time. Other statements can be found in the appendix.

Despite the large variety of locus of control surveys it was chosen to include Rotter’s locus of control survey. This decision was made because all other surveys existed out of more than thirty questions. Because subjects already participated in a real effort task for three minutes there was the fear that subjects would not read the questions carefully anymore because of boredom

Figure 2: Summary of treatments

C. HYPOTHESES

The main interest of the current study is to test whether (1) different giving behaviour is observed in case a positive external shock is made salient, and (2) whether the same giving behaviour is observed when a positive external shock is present but not received as when income is determined by effort only. Furthermore, this study investigates whether beliefs about luck influence giving behaviour.

To test these hypotheses two treatments were designed; a control treatment and a bonus treatment. The control treatment mainly served to check whether the introduction of a bonus changed giving behaviour in case subjects did not receive the bonus. The bonus treatment served to verify whether a positive external shock would influence giving behaviour.

Drawing on the literature discussed in the previous chapter the following predictions can be derived. Based on similarity of the study of Sheremeta et al. (2011) it is predicted that

(17)

17

subjects are willing to give more when they received the bonus. To verify these hypotheses, the following null and alternative hypotheses are specified:

H0: The average amount given to the matched participant is not influenced by a

positive external shock, i.e.

𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡,𝑛𝑜 𝑏𝑜𝑛𝑢𝑠}= 𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡, 𝑏𝑜𝑛𝑢𝑠}

H1: The average amount given to the matched participant increases when the final

score is affected by a positive external shock, i.e.

𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡, 𝑛𝑜 𝑏𝑜𝑛𝑢𝑠}< 𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡, 𝑏𝑜𝑛𝑢𝑠}

The second prediction is that giving behaviour does not differ in the control group from the treatment group when subjects do not receive the bonus. The reason for this prediction is that in past research it is found that when property and entitlements rights come in to play this tend to causes self-regarding behaviour (Hoffman et al, 1994). The fact that the other person is lucky in the bonus treatment should therefore not change behaviour in the dictator game. To test these predictions, the following null and alternative hypotheses are specified:

H0: The average amount given to the matched participant is the same in the control

group and in case of absence of the bonus treatment group, i.e.

𝜇_{𝑔𝑖𝑣𝑒𝑐𝑜𝑛𝑡𝑟𝑜𝑙}= 𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡, 𝑛𝑜 𝑏𝑜𝑛𝑢𝑠}

H1: The average amount given to the matched participant differs in the control group

from the case of absence of the bonus treatment group, i.e.

𝜇_{𝑔𝑖𝑣𝑒𝑐𝑜𝑛𝑡𝑟𝑜𝑙}≠ 𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡, 𝑛𝑜 𝑏𝑜𝑛𝑢𝑠}

Regarding the beliefs about luck it is predicted that subjects with an external locus of control are giving more than subjects with an internal locus of control. It is expected that giving behaviour of externals is comparable with experimental results in which subjects get a windfall of money. Giving behaviour of internals is predicted to be more in line with results from real effort experiments. It is found that in real effort experiments subjects tend to give less than when they got a windfall of money (Cherry et al, 2002). To test these predictions, the following null and alternative hypotheses are specified:

(18)

18

H0: The average amount given is the same for subjects with internal locus of control

beliefs as for subjects with external locus of control beliefs, i.e.

𝜇_{𝑔𝑖𝑣𝑒𝑐𝑜𝑛𝑡𝑟𝑜𝑙, 𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙} = 𝜇_{𝑔𝑖𝑣𝑒𝑐𝑜𝑛𝑡𝑟𝑜𝑙, 𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙}

𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑛𝑜 𝑏𝑜𝑛𝑢𝑠), 𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙} = 𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑛𝑜 𝑏𝑜𝑛𝑢𝑠), 𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙}

𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑏𝑜𝑛𝑢𝑠), 𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙} = 𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑏𝑜𝑛𝑢𝑠), 𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙}

H1: The average amount given smaller for subjects with internal locus of control

beliefs than for subjects with external locus of control beliefs, i.e. 𝜇_{𝑔𝑖𝑣𝑒𝑐𝑜𝑛𝑡𝑟𝑜𝑙, 𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙} < 𝜇_{𝑔𝑖𝑣𝑒𝑐𝑜𝑛𝑡𝑟𝑜𝑙, 𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙}

𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑛𝑜 𝑏𝑜𝑛𝑢𝑠), 𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙} < 𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑛𝑜 𝑏𝑜𝑛𝑢𝑠), 𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙}

𝜇_{𝑔𝑖𝑣𝑒𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑏𝑜𝑛𝑢𝑠), 𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙} < 𝜇_{𝑔𝑖𝑣𝑒𝑐𝑜𝑛𝑡𝑟𝑜𝑙, 𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙}

4. RESULTS

This chapter presents the results from the online conducted experiment. The first part provides some descriptive statistics to ensure that the control group and treatment group do not differ in terms of demographics and performance. The second part analyses giving behaviour of both treatments and makes a comparison between the two treatments. This chapter concludes with a section about how the beliefs about luck influenced giving behaviour.

A. DESCRIPTIVE STATISTICS

As mentioned before this experiment has been conducted online, making use of the online CREED website. A total of 88 subjects participated (n = 88), of which 46 were randomly assigned to the control treatment and 42 were randomly assigned to the bonus treatment. One subject was excluded from the experiment, because he worked six instead of three minutes on the real effort task.

Table 1 provides some additional insight about the population demographics. A Mann-Whitney non-parametric test was conducted to check whether any population difference would arise. As can be seen, it is not possible to reject the null hypothesis that both population demographics are the same for age, gender and schooling. The only demographic

(19)

19

variable that seems to differ between the two populations is income. This could result in more self-regarding behaviour in the bonus treatment, because they might need the money more.

There do not seem to be any differences between the groups when it comes to experience in economic experiments. Because this might influence behaviour (Hoffman et al, 1996), when comparing both populations it is important that this does not differ.

Regarding the real effort task, there do not seem to be any differences either. When comparing the level of the task (ranging from “very easy” to “very hard”), the level of effort (ranging from “not hard” to “very hard”), the number of puzzles solved correct and the number of puzzles solved wrong, the Mann-Whitney non-parametric test cannot reject the null hypothesis. This evidence seems to rule out the possibility that different observed giving behaviour could be due to the first stage of the experiment.

Table 1: Summary statistics on population demographics and performance

Note: * indicates statistical significance at 10% level, ** at 5% level, and *** at 1% level. Standard errors are in the parentheses.

Treatment Control Bonus Mann-Whitney test

Mean Min Max Mean Min Max p-value

Age 25.76 (6.98) 19 52 24.90 (6.74) 16 51 0.35 Female 0.63 (0.49) 0 1 0.55 (0.50) 0 1 0.43 Schooling 5.08 (1.13) 1 5 4.64 (1.57) 0 1 0.33 Income 2.35 (0.70) 1 3 2 (0.80) 1 3 0.04** Experiment before 0.48 (0.50) 0 1 0.62 (0.49) 0 1 0.19 Level task 2.76 (0.84) 1 5 2.76 (0.73) 1 5 0.74 Effort 3.13 (1.09) 1 5 3.21 (1.12) 0 5 0.44 Correct 4.78 (1.82) 0 9 5.02 (2.07) 1 9 0.40 Wrong 2.23 (1.71) 0 8 1.92 (1.42) 0 5 0.41 LOC 6.07 (2.24) 1 12 6.21 (2.04) 1 11 0.79 Feeling bonus - - - 3.02 (0.56) 1 4 -

(20)

20

B. DETERMINANTS OF GIVING

This section discusses the determinants of observed giving behaviour. First, the first two hypotheses are tested. Second, an ordinary least squares regression is performed to test the predictive power of both demographic and performance variables.

I. WILCOXON MATCHED-PAIR TEST & MANN-WHITNEY TEST

Because the demographic and performance measures of both populations seem to be equal, it is possible to compare both groups. Table 2 shows the average transfers made in both treatments. The third column shows the average amount given for the control group and for the treatment group in case people did not receive a bonus but still had the highest final score. The fourth column only shows the average transfer in the hypothetical situation subjects had the highest final score and received the bonus.

To check the first hypothesis a Wilcoxon matched-pair signed-rank test was performed. When the average amount given in case subjects received no bonus was compared with the average amount given when subjects received a bonus a p-value of 0.02 was found. This means that based on these experimental findings, with a significance level of 5% the null hypothesis can be rejected. In other words, giving behaviour tends to differ when subjects receive a positive external shock from the case in which they do not.

To test whether the existence of a bonus influenced giving behaviour a Mann-Whitney test was conducted. When comparing the control group with the treatment group in case subjects did not receive the bonus a p-value of 0.70 was found. This means that there is not enough evidence to reject the null hypothesis which stated that giving behaviour does not differ when a bonus is present but not received. Based on this experimental finding it thus can be stated that when a bonus is present and subjects still earn the entitlements to make a decision about the division of the money, giving behaviour does not seem to differ.

Figure 3 shows the cumulative distribution of the transfers made. As can be seen, giving behaviour in the control group and the treatment group without bonus seems to be equal. With only 40% of the subjects offering zero instead of 60%, giving behaviour in the treatment group with bonus seems to be different.

Subjects were also asked on what kind of intuition they based their giving behaviour. Because this was an open question, the answers are categorized in the following categories: Anonymity, which includes reasons as “because I do not know the other participant”, fairness i.e. “fair amount” and “giving the other person one euro for the effort”, greed; including all

(21)

21

self-regarding behaviour, property which includes reasons as “I worked hard and wanted all my earnings”, equality which were persons who wanted outcomes to be equal and other reasons which cannot be classified under the previous mentioned categories. The results of this question can be found in Figure 4. Because subjects in the treatment group were asked to make a decision about what to give to the other participant in case they won and did not receive the bonus, and in case when they won and did receive the bonus, a distinction is made between three groups; Give Control, Give No Bonus (which refers to the situation in which subjects won and did not receive a bonus) and Give Bonus (which refers to the situation in which subjects won and did receive a bonus).

Table 2: Transfers made in both treatments

Standard errors are in the parentheses.

Figure 3: Cumulative distribution of transfers

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 0 1 2 3 4 5 6 7 8 9 10 C umul at ive F re qu ency (% ) Amount given Give control Give no bonus Give bonus Treatment Number of observations Average transfer (no bonus) Average transfer (bonus) Wilcoxon matched-pair signed-rank test (p-value) Control 46 1.39 (2.07) - Bonus 42 1.54 (2.11) 2.33 (2.48) 0.02 Mann-Whitney test (p-value) 0.70

(22)

22

The reasons provided seem to differ between these three groups (Figure 4). For example, while in the control group fairness is given as reason for giving behaviour in approximately 15% of the time, it is only provided in 7% of the time in the treatment group when the other participant received the bonus. Furthermore, property is only mentioned 16% of the time in case of the control group, in the no bonus group it is stated to be the reason in almost 30% of the time. This enhanced effect of property rights could be due to the fact that in case of the no bonus group, subjects were asked what they would do if they won while the other participant received the bonus. Because the bonus was perceived too large by many subjects (which is further deliberated in the discussion), the possibility that their matched participant still lost while receiving the bonus, might have been a signal of really low effort. The proportion of self-regarding subjects seems to be equal in both groups.

When comparing give no bonus and give bonus another interesting finding is revealed. Although property is mentioned in 30% of the time in the hypothetical situation subjects won without receiving the bonus, this amount decreases sharply in the case subjects did receive the bonus to almost 5%. Furthermore, fairness reasons sharply increased from 7% in the No Bonus case, to almost 30% in the case subjects received the bonus. As an example, in case of the bonus one of the fairness reasons provides was: “There was any help, so I could give something to the other.”

Additionally, when comparing the three groups a different level of egalitarians can be observed in Give control. This may explain why in case of ‘no bonus’ the amount transferred is a bit higher compared to the control group. What is interesting to note, is that the egalitarians in the treatment group are not affected by the bonus; if they decided to equalize earnings in the case they did not receive the bonus they did it in the exact same way when they received the bonus.

II. OLS

To check whether giving behaviour in both treatments could be predicted by demographic and performance variables an OLS regression was performed. The results of the regression can be seen in Table 3. As can be seen all regressions have a very low R-squared. This means that the predictive power of all models is poor.

Despite the low R-squared this regression reveals that when subjects are higher educated they tend to give less. This result seems to be significant when combining the

(23)

23

control and no bonus group, and even highly significant when final earnings are influenced by a positive external shock.

Female subjects tend to give more than male subjects, which is consistent with the previous discussed literature (Adreoni & Vesterlund, 2001). This finding seems to be only significant in the case of no bonus. Furthermore, income seems to have a positive effect on giving. This could be due to the fact that subjects only could earn a maximum of ten euros, which might have been not influential enough for “rich” subjects.

When it comes to experimental experience, the fact that subjects participated in an experiment before tends to have a negative effect in the control group and in the case subjects did not receive a bonus. Surprisingly, this effect seems to be positive when subjects did receive the bonus. Because all these findings are not significant it is not possible to draw conclusions from this results.

Finally, looking at the effects of the real effort task the results of how subjects perceived the level of the task differ across treatments. In the control group the perceived level seems to have a negative effect on giving, while in the treatment group a positive effect is found. This suggests that in the control group subjects were giving less the harder they perceived the task. In the treatment group the reverse effect is observed, the harder the task, the more they were willing to give.

Also the effects of effort are not consistent across both groups. They seem to be positive in the control group, while they seem to be negative in the bonus group. This suggest that in the control group subjects who worked harder were more willing to give away part of their earnings, while in the treatment group the reverse effect was found.

The amount of correct answers tends to have a negative effect over all treatments. Because the results found regarding the level of the task, the exerted effort and the correct answers are all not significantly different from zero, it is hard to tell whether these effects will be found in other experiments as well.

Subjects were also asked about how they felt after they got the information about the external positive shock. Surprisingly subjects who indicated to be angry after being told about the bonus tend to give more than subjects who indicated to be happy. Because these findings are not significant it could be the case that in other experiments different results will be found.

(24)

24

Figure 4: Reasons given for giving behaviour

Table 3: OLS regression using demographic and performance variables

Treatment Give

(Control & No Bonus)

Give Control

Give No Bonus Give Bonus

Age 0.04 (0.04) -0.06 (0.05) 0.13 (0.05) *** 0.06 (0.06) Female 0.74 (0.44) * 0.48 (0.66) 1.39 (0.59) ** 0.71 (0.73) Schooling a _{-1.61 (0.62) *} _{-1.94 (1.22)} _{-1.96 (0.81) **} _{-3.45 (1.00) ***} Income 0.19 (0.30) 0.40 (0.48) 0.10 (0.40) 0.73 (0.50) Experiment before -0.53 (0.45) -0.66 (0.62) -0.24 (0.64) 0.60 (0.79) Level Task 0.04 (0.30) -0. 93 (0.51) 0.44 (0.42) 0.70 (0.52) Effort -0.20 (0.20) 0.30 (0.34) -0.40 (0.27) -0.20 (0.33) Correct -0.10 (0.12) -0.30 (0.21) 0.01 (0.16) 0.00 (0.20) Feeling bonus b _- _- _{0.76 (0.80)} _{0.91 (1.00)} Constant 2.13 (1.77) 6.91 (3.52) * -1.10 (2.32) -0.07 (2.88) R-squared 0.2003 0.1875 0.4906 0.4324 Adjusted R-squared 0.1193 0.0118 0.3473 0.2727

Note: * indicates statistical significance at 10% level, ** at 5% level, and *** at 1% level. Standard errors are in the parentheses.

a_{Because schooling is a nominal variable, for the OLS regression a binary variable was created. In this case 1}

means higher educated, which includes having obtained a degree at the University of Applied Science, a Bachelor degree and a Master degree

b_{Binary variable which is 1 in case subjects indicated to be angry or very angry}

0% 5% 10% 15% 20% 25% 30% 35% R eas on provi ded (% ) Give control Give no bonus Give bonus

(25)

25

C. BELIEFS ABOUT LUCK

At the end of the survey subjects had to fill in a locus of control questionnaire. This was a questionnaire based on Rotter’s locus of control survey, which existed of thirteen questions. Based on the given answers a locus of control measure was determined ranging from one, which is the extreme case of internal control beliefs, to thirteen, which is the extreme case of external control beliefs. Furthermore, one question about how subjects perceived luck related to work was taken out to check whether these beliefs about how effort relates to success would influence giving behaviour. This variable is a binary variable which is one in case of external control beliefs.

When comparing these measures between the two groups the Mann-Whitney test fails to reject the null hypotheses that both groups tend to have different beliefs (Table 1). Furthermore, one can see that according to the mean of the locus of control measure subjects tend to have average beliefs, so not extremely internal or external.

Table 4 shows the correlation between the measure of the locus of control and observed giving behaviour in the control group. The correlation of the measure of locus of control seems to be positive in case of the control group and negative in the treatment group. Because the correlation between the locus of control and giving behaviour in the control group is significant at a 5% level only, it could be the case that the negative relationship found in the treatment group is not present.

Regarding subjects’ perceptions about how success is determined it seems to be the case that in the control group external control beliefs were negatively correlated with giving behaviour, while in the treatment group external control beliefs tend to be positively correlated with the amount given. Nevertheless, all correlations regarding beliefs about success are highly insignificant. This means that there is probably no relationship between observed giving behaviour and perceptions about how success is obtained.

To test whether subjects with an external locus of control tend to give more to others than subjects with an internal locus of control, a non-parametric Mann-Whitney test was conducted. Because this test only allows to make a distinction between two groups, subjects were classified either as having internal beliefs or external beliefs (Table 5). Subjects with a low locus of control measures (ranging from one to six) were classified as internals and subjects with high locus of control values (ranging from seven to thirteen) were classified as externals.

(26)

26

When conducting a Mann-Whitney test it is found that in the control group giving behaviour seems to differ between internals and externals. This does not seem to be the case in the treatment group. In the control group a p-value of 0.03 is found, which means the null hypothesis can be rejected on a 5% significance level. This suggests that in the control group giving behaviour differs between subjects with an internal locus of control and subjects with an external locus of control.

These differences are not found in the treatment group. With the p-values of respectively 0.68 and 1 it is not possible to reject the null hypothesis. This suggests that in the treatment group giving behaviour does not seem to differ between subjects with an internal locus of control and subjects with an external locus of control. Therefore, the hypothesis which states that people with internal locus of control beliefs give the same as people with external locus of control beliefs can only be rejected in case of the control group.

Table 4: Correlations giving and locus of control

Give Control Give No Bonus Give Bonus Locus of Control Locus of Control 0.29 (0.05) -0.13 (0.41) -0.14 (0.36) -

Locus of Control (Work)

-0.06 (0.69) 0.02 (0.89) 0.07 (0.65) 0.46 (0.00)

p-values are in the parentheses.

Table 5: Average transfers ordered by locus of control

Number of observations Average transfer (no bonus) Average transfer (bonus) Control group - Internal beliefs 25 0.68 (1.44) - External beliefs 21 2.24 (2.41) - Mann-Whitney test (p-value) - 0.03 - Treatment group Internal beliefs 23 1.74 (2.20) 2.30 (2.40) External beliefs 19 1.32 (2.03) 2.37 (2.63) Mann-Whitney test (p-value) 0.68 1

(27)

27

5. DISCUSSION

In the current study three predictions were made. First, it was predicted that when subjects were positively affected by an external shock, this would increase the amount given compared to the situation in which they did not receive a bonus (Sheremeta et al., 2011). Secondly, it was predicted that giving behaviour would be the same in the control group as in the treatment group in case subjects did not receive a bonus (Hoffman et al, 1996). Finally, it was predicted that people with an internal locus of control would give less than people with an external locus of control.

The results from this experiment seem to be consistent with the predictions made about the effect of luck when it is made salient. This offers several valuable lessons. Governments may use these findings when promoting their redistribution behaviour. Making the higher income individuals aware of the luck they might have had could convince them to show more generous behaviour. Furthermore, charities could use these findings when recruiting additional financial donors. Highlighting the luck people in wealthy economic areas had, may persuade them to give something to the poor.

Nevertheless, there are several considerations that are worthwhile to point out. First the internal shortcomings are discussed. To begin with, a shortcoming of the experiment is that it has been conducted online. Instead of conducting an experiment in the lab, this results in a lack of control of external factors (like noise). Furthermore, it does not allow subjects to ask questions when they do not understand what they have to do. This might have had an influence on the results. The answers given to the question “How did you decide what to give in case you did (not) receive the bonus?” revealed that a lot of subjects did not understand how the bonus worked out. According to the answers, some subjects thought that the bonus was in terms of money instead of points. Furthermore, some answers revealed that is was not made clear enough that the decisions subjects made would only be implemented in case they had the highest final score.

Secondly, conducting the experiment online had one other drawback; because subjects were not participating in the experiment at the same time, they had to make decisions about hypothetical situations. This means that subjects may have acted according how they would ideally behave, and not how they would behave if something like this would happen in real life. This might have reduced the internal validity of the study.

Thirdly, a shortcoming in terms of the experimental design is that the external positive shock might have been too high. Before the experiment was published online, several test

(28)

28

rounds were performed to find the average performance of the real effort task and an average result of six puzzles was found. When analysing the data of the experiment, it was found that the average score of the control group was only 4.7 and the average score of the bonus group was 5. As the external positive effect was meant to be half of the average score, the external positive shock was too high in this experiment. Several subjects also mentioned this.

Fourthly, another shortcoming is that only one pair of participants was selected for payment. Because using a payment structure to induce monetary value on decisions is an essential element of experimental control (Smith, 1976), only selecting one pair for payment may have reduced control in this experiment. This could have resulted in more generous offers, while subjects may have made less generous offers when all would have been paid.

Besides these internal shortcomings, the results might also suffer from several external considerations. Firstly, although the design of the experiment was meant to be as simple and abstract as possible, one might be questioning its applications to real life. External positive effects might be present when looking at one’s achievements in life, but it probably will never be as clear cut as in this experiment. This might give some room for the self-serving bias to come into play, and as Tonin and Vlassopoulos (2016) found, this tends to reduce the positive effect on generosity found in this study.

Secondly, situations tend to be never as structured as in this study – where only two people interact and only one receives an external positive shock. First, most people tend to interact with more than one person.

Finally, it might even be the case that in real life everyone does receive a positive external shock at one point in life but that only the height of this external shock really influences behaviour.

Future research should take the previous discussed shortcomings into consideration. This means that the experiment should be conducted in a lab to control for external effects and to make sure all subjects understand the instructions, the height of the external positive shock should better suit to the average performance and it should have a different payment structure, where instead of one pair, all subjects are paid.

In addition, it could be interesting to run the same experiment but then with two different heights of the positive external shock. This allows to test whether the height of the external shock does affects generosity, or that giving behaviour is only affected by the appearance of a positive external shock.

(29)

29

6. CONCLUSION

This study answers the research question whether the saliency of a positive external shock does influence generosity. In other words, this study tried to find whether explicit awareness of luck affects giving behaviour in such a way that subject care more about fairness. Drawing on related literature three predictions were made. First, it was predicted that a positive external shock would change giving behaviour, in a way that subjects would be more generous when they received the bonus. Second, it was predicted that when a positive external shock was present but not received, giving behaviour would not change. Third, it was predicted that people with an internal locus of control would give less than people with an external locus of control. To test these hypotheses an online experiment has been conducted.

Based on the results, the following conclusions can be drawn. According to several tests, in terms of demographic and performance variables there did not seem to be notable difference between the two treatment groups. Furthermore, a rejection of the first null hypothesis was possible, which means that when subjects are positively affected by luck, they tend to show more generous behaviour. In addition, it was not possible to reject the null hypothesis which stated that giving behaviour in the control group is the same as giving behaviour in the treatment group when subjects did not receive the bonus. This is consistent with the predictions made. This means that these findings tend to correspond with the predictions. In case of the beliefs about luck it was only possible to reject the null hypothesis in the control group. In the treatment group there did not seem to be any differences in the amount given between subjects with internal and external locus of control beliefs.

This results could be interpreted in two ways. First, it could be the case that people tend to care more about fairness when they are positively affected by luck. Second, it could be the case that fairness preferences change when luck is made salient. The importance of the findings of this study should not be underestimated; the fact that subjects tend to care more about fairness when they are positively affected by luck (or have different fairness preferences when luck is made salient) could help governments with promoting redistribution. Additionally, charity funds may as well use these findings in their campaign to recruit additional financial donors.

Although some findings of this experiment are consistent with the predictions, one should be careful when drawing on these conclusions, because this experiment suffered from some shortcomings. This includes a lack of control of external factors and a lack of control of

(30)

30

making sure that subjects understood everything in the experiment, the height of the positive external shock and the payment structure. Future research could strengthen the findings by improving these shortcomings and by investigating the effect of the height of a positive external shock.

(31)

31

BIBLIOGRAPHY

Andreoni, J. (1989). Giving with impure altruism: applications to charity and Ricardian equivalence. The Journal of Political Economy, 1447-1458.

Andreoni, J., & Vesterlund, L. (2001). Which is the fair sex? Gender differences in altruism. Quarterly Journal of Economics, 293-312.

Auten, G. E., Sieg, H., & Clotfelter, C. T. (2002). Charitable giving, income, and taxes: an analysis of panel data. The American Economic Review, 92(1), 371-382.

Becker, A. (2013). Accountability and the fairness bias: the effects of effort vs. luck. Social Choice and Welfare, 41(3), 685-699.

Buckley, E., & Croson, R. (2006). Income and wealth heterogeneity in the voluntary provision of linear public goods. Journal of Public Economics, 90(4), 935-955. Camerer, C. (2003). Behavioural game theory: Experiments in strategic interaction.

Princeton University Press.

Carpenter, J., Liati, A., & Vickery, B. (2006). They come to play: supply effects in an economic experiment. Middlebury College, Department of Economics WP,(0602). Cherry, T. L., Frykblom, P., & Shogren, J. F. (2002). Hardnose the dictator. The American

Economic Review,92(4), 1218-1221.

Côté, S., House, J., & Willer, R. (2015). High economic inequality leads higher-income individuals to be less generous. Proceedings of the National Academy of Sciences, 112(52), 15838-15843.

Eckel, C. C., De Oliveira, A., & Grossman, P. J. (2007). Is more information always better? An experimental study of charitable giving and Hurricane Katrina. Southern

Economic Journal, 74(2).

Erkal, N., Gangadharan, L., & Nikiforakis, N. (2011). Relative earnings and giving in a real- effort experiment. The American Economic Review, 101(7), 3330-3348.

Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. Quarterly journal of Economics, 817-868.

Findley, M. J., & Cooper, H. M. (1983). Locus of control and academic achievement: a literature review. Journal of personality and social psychology,44(2), 419. Fong, C. M. (2007). Evidence from an experiment on charity to welfare recipients:

Reciprocity, altruism and the empathic responsiveness hypothesis. The Economic Journal, 117(522), 1008-1024.

Furnham, A. (1986). Economic locus of control. Human relations, 39(1), 29-43.

Glazer, A., & Konrad, K. A. (1996). A signaling explanation for charity. The American Economic Review, 86(4), 1019-1028.

Hoffman, E., & Spitzer, M. L. (1985). Entitlements, rights, and fairness: An experimental examination of subjects' concepts of distributive justice. The Journal of Legal Studies, 14(2), 259-297.

Hoffman, E., McCabe, K., Shachat, K., & Smith, V. (1994). Preferences, property rights, and anonymity in bargaining games. Games and Economic Behaviour, 7(3), 346-380. Hoffman, E., McCabe, K., & Smith, V. L. (1996). Social distance and other-regarding

behaviour in dictator games. The American Economic Review, 86(3), 653-660. Kestenbaum, J. M., & Hammersla, J. (1976). Filler items and social desirability in Rotter's

The lucky one : giving behaviour in a real effort experiment