Thesis bachelor of economics
Provision of an optional public good and the incentives
for non-cooperative behaviour
Abstract
This thesis set out to determine the provision of an optional public good through the use of a
public good experiment. It is found that the provision is similar, but not identical, to that in
standard public good experiments. The difference being that some participant choose the
loner strategy over the free rider strategy. This is observed in all rounds of the experiment,
although it is more significant in the final rounds. Furthermore, it is reasoned that the loner
strategy will be chosen by participants that invest in the private good for reasons besides profit
maximization. Therefore, it can be concluded that some contributors are pushed towards the
free rider strategy in non-optional public good experiments.
Name:
Joppe Penning
Student number:
10424679
Programme:
BSc Future Planet Studies
Statement of Originality
This document is written by Student Joppe Penning 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.
Introduction
The voluntary provision of a public good can be linked to several societal issues and the role that markets might have in solving them (Ledyard, 1995). Some examples are: causes and effects of climate change, public health care, environmental concerns and flood protection. These complex societal issues are of such importance that they have to be dealt with one way or another. Provision through voluntary contributions is one of the options to try to provide the social optimal allocation. This is where public good experiments come in as they have a key role in finding out to what level and under what circumstances people are willing to cooperate.
In current public good literature, public goods are usually considered to be non-optional (Chaudhuri, 2011; Ledyard, 1995; Zelmer, 2003), meaning that everyone benefits whether they are aware of it or not. This seems to be representative for some public goods like dykes, carbon emission reduction programs and street lightning. However, there are also many public goods for which this does not hold. Those goods are optional public goods and they directly benefit the users only. Parks, museums, websites and radio channels are examples that fit in this category. It has to be noted that parks and museums are more common than public goods, due to their slight rival character, whereas radio channels and websites are fitting the description of a public good. They are rivalrous and non-excludable. As optional public goods do not benefit everyone, non-users have an incentive not to contribute. Therefore, this thesis sets out to determine whether the provision of an optional public good differs from the provision of a non-optional public good.
Furthermore, optional public good experiments can be helpful in determining the incentives of participants not to contribute to the public good. In non-optional public good experiments a participant can be a contributor, a free rider or more commonly a combination of the two (Fischbacher, Gächter, & Fehr, 2001). However, in optional public good experiments participants can also choose the loner or the altruist strategy. The addition of the loner strategy removes unintentional and unwanted free riding by participants whom withhold from contributing to the public good. Hence, the participants that choose this strategy, do not contribute to the public good for reasons different from trying to maximize profit.
This thesis is set up as follows. Firstly, the current perspective on public good provision is discussed in the theoretical framework. Next, the choices in and the implementation of the optional and non-optional public good experiment are elaborated upon. Thereafter, the results of both experiments, and the conclusions drawn from them, will be discussed. Lastly, the flaws of this study as well as recommendations for further studies are brought forward.
Theoretical Framework
The description of a public good experiment given by Ledyard (1995) is often used as the benchmark to explain how a public good experiment works (Andreoni, 1995; Chaudhuri, 2011; Croson, 1996; Keser & Van Winden, 2000). I will, therefore, provide an abstract version of his explanation to illustrate the basics mechanics of a public good experiment.
In a public good experiment, a number of n participants forms a group for the duration of x rounds. The participants are not allowed to communicate during the experiment. In each round, every participant receives an endowment of ω. Subsequently, each participant simultaneously has to make a personal contribution of C (0 ≤ 𝐶 ≤ 𝜔) to the public good for that particularly round. Their own
contribution as well as the contribution of the other participants determines what their eventual payment is.
The payment per round of each participant consists out of the private earnings (= 𝜔 − 𝐶) and the public good yields, which is equal to the marginal per capita return (MPCR) α (0 ≤ 𝛼 < 1) times the total contribution to the public good. At the end of each round the participants are informed what their earnings of the previous round were. Subsequently, the participants have to decide how much to invest in the public good for the next round. These steps will be repeated until x rounds have been played.
The social optimal outcome in public good experiments is for all participants is to contribute their full endowment. However, experimental evidence shows that the social optimal is not achieved through voluntary contribution (Chaudhuri, 2011).
Variables
Since the first contribution to the theory on the voluntary provision of a public good by Olson (1965), there has been a growing interest in the provision of such goods without any form of government intervention (Bohm, 1972; Sweeney, 1973). However, it was only in the eighties that the effect of various variables were studied in experiments (Andreoni, 1988; Isaac & Walker, 1988; Kim & Walker, 1984). Since then it has been established that the level of control in public good experiments determines the level of contribution to a public good (Ledyard, 1995). Therefore, replacing the non-optional public good by an non-optional public good will add to the knowledge on the provision of public goods.
As stated before, the level of control and the choice of variables has a strong effect on the provision of a public good. Therefore, the numerous experiments on public good provision usually choose one or more of the variables to focus on. Andreoni (1988) studied the effect of participants randomly changing groups such that their group compensation was different in every round of the experiment. This enabled him to make a distinction between strategic behaviour and learning by the participants in his experiment. Isaac & Walker (1988) focused on two completely different variables. Their main interests were the effects of group size and the value of the marginal per capita return. The results of both studies indicate that changing one or more of the variables has a significant effect on the outcome of the experiment. Next to these variables, many other variables that affect the level of cooperation during public good experiments have been identified (Kim & Walker, 1984). Therefore, the level of control has to be taken into account when comparing public good experiments.
Free rider problem
There are two conflicting interest at play in public good experiments. On the one hand, there is self-interest which works as an incentive to invest in the private good. On the other hand, the self-interest of the group dictates an investment in the public good. Thus, where any investment in the private good maximizes personal profits, an investment in the public good maximizes total welfare. These opposite interests create tension, especially since the participants know that each of the other participants is confronted with the same dilemma.
The perspective of neo-classical economists and game theorists on public good experiments is as follows. They argue that since a public good game is played a finite amount of times, all participants will choose to free ride. This conclusion arises from the application of backwards induction. In the final round the profit maximizing decision is to free ride and contribute nothing to the public good. There will not be any further rounds, thus others cannot react to your free ride behaviour. However, since all
participants know this, free riding in the second to last round is also the profit maximizing choice. This train of thought can be applied up to the first round. Hence, all participants will free ride in all of the rounds.
The free rider problem was immediately connected to the provision of a public good (Olson, 1965). However, in the late seventies and early eighties, some studies found that free riding occurred less than expected or not at all (Kim & Walker, 1984). Bohm (1972), for example, found that subjects did not understate their willingness to pay to watch a television series, while knowing that their stated values would be their cost.
However, the outcome of these early experiments in the realm of public good provision do not match with the current view on free ride behaviour during public good experiments (Chaudhuri, 2011). Later studies found that in the first round of a public good experiment the contribution to the public good game is on average between the 40 and 60 percent (Andreoni, 1988; Croson, 1996; Isaac & Walker, 1988; Keser & Van Winden, 2000; Weimann, 1994). This is in contradiction with the economic and game-theoretic view on public good provision. Therefore, public experiments are an irreplaceable tool to find out how to what extend people cooperate to achieve the social optimal provision of a public good (Ledyard, 1995).
The level of cooperation decreases as the experiment approaches the end. In the final round of a ten round public good experiment the endowment to the public good has decreased to about 5 to 20 percent (Andreoni, 1988; Isaac & Walker, 1988). However, strict free riding, which is the dominant profit maximizing strategy, by all of the participants is seldom observed (Weimann, 1994).
Punishment
Punishment of free riders is an effective tool to maintain cooperation in public good experiments (Fehr & Gächter, 1999). However, in most public good experiments the contribution to the public good is anonymous. Therefore, contributors can only punish free riders by stopping to invest in the public good. Andreoni (1995) argues that this, next to learning the profit maximizing strategy, is one of the main reasons of decaying contributions. Moreover, it should be acknowledged that in standard public good experiments, default contributors become free riders themselves when they switch to investing in the private good as they automatically make use of the public good.
Partners and strangers
One of the variables that strongly affects free riding behaviour is the persistence of the group composition (Keser & Van Winden, 2000). The theory is as follows. When participants are assigned to a fixed group they have an incentive to strategically contribute to the public good. Free ride behaviour will namely deter others from contributing in later rounds as it indirectly educates others about the profit maximizing strategy. However, when the group composition randomly changes from round to round, the incentive to contribute strategically to the public good disappears. Therefore, the contributions of the stranger group will decay faster than the contributions of the partner group. Andreoni (1988) was the first to study this theory in his strangers and partners experiments. Although, he found results that were not in accordance with the theory. Throughout all rounds the strangers contributed more to the public good and less were free riding than the partners. Both groups did, however, approach the free riding level towards the final round. Therefore it can be concluded that participants do learn the profit maximizing strategy over the course of the experiment. Croson (1996) replicated the strangers and partners study of Andreoni (1988) and did find results in accordance to the theory. Moreover, a similar study by Keser & Van Winden (2000) also found results that confirmed
the theory. Hence, the theory is currently considered to be valid. Furthermore, this displays the importance of repeating economic experiments.
Optional public good
In current literature there is no mentioning of optional public good experiments. However, optional public good experiments can be a vital tool in understanding the motivation of participants to cooperate or to free ride. Firstly, I will discuss the different strategies and thereto bound incentives in an optional public good experiment. Since there are two decisions to be made in an optional public good experiment, there is a total of four strategies. These are the free rider, contributor, loner and altruist strategies. The free rider and contributor strategies are also present in standard public good experiment, but the loner and altruist strategies are specific to optional public good experiments. The decisions of the different strategies are displayed in table 1 below.
Public good use
Yes No
Investment
Private good Free rider Loner
Public good Contributor Altruist Table 1. Overview Strategies Optional Public Good Experiment
The free rider strategy is chosen by participants that want to maximize their profits in the current round. This strategy namely yields the highest per round earnings. The contributor strategy is already slightly more refined as there are two incentives to choose this strategy. The first possible motivation that induces cooperation is altruism. When a participant values total welfare more than personal profit, this strategy will be preferred over the free rider strategy. Participants that choose the altruist strategy are also driven by this incentive. Secondly, strategic behaviour can be a reason to become a contributor as participants might expect that a cooperative behaviour will induce others to also contribute to the public good in subsequent rounds. Furthermore, free riding might work as an indirect educational tool for showing others the profit maximizing strategy (Andreoni, 1988).
The choice for the loner strategy can arise from the expectation that the contributor strategy will yield a lower profit. This would be the case if all of the others decide to invest in the private good. The maximum profit by investing in the public good is then lower than the profit gained from an investment in the private good. Put differently, choosing the contributor strategy yields, as always, a lower profit than the free rider strategy, but also yields a lower profit than the loner strategy. A participant might expect this to be the case in the final round since the free rider strategy is the overall profit maximizing strategy in that round.
Another reason to choose the loner strategy can be the will to punish free ride behaviour by others. Since punishment is not included in the optional public good game, the only way to decrease the profit of free riders is to stop investing in the public good. Thereby it is important to acknowledge that a participant, whom invests in the private good, would become a free rider him or herself by using the public good. This is the exact behaviour that these type of participants try to punish by investing in the
private good. Therefore, these participants will most likely choose the loner strategy over the free rider strategy. Punishing free riders could be one of the most important incentives for participants to choose the loner strategy as it is shown that free riders are heavily punished in public good experiments (Fehr & Gächter, 1999).
Methodology
In this section the design and the implementation of the experiments is discussed. There are two main takes on public good experiments. The difference lies in the calculation of the value of the public good. In one version, the total contribution is multiplied by a constant factor 𝛼 (0 < 𝛼 < 1), while in the other one the total contribution is multiplied by 𝛽 (> 1) and then divided by the number of users. The first version uses the payment structure of a pure public good, while the second one is more applicable to a common good. The prevailing version seems to be the one with a MPCR (0 < ∝ < 1) (Andreoni, 1988; Chaudhuri, 2011; Ledyard, 1995; Zelmer, 2003) and will therefore be used in our experiments. The provision of optional public goods has been researched using an experiment, based on the study of Andreoni (1988). His study uses a standard public good experiment as described by Ledyard (1995). Eventually, two different experiments were conducted. The first one studies the provision of an optional public good, while the second one functions as a base line experiment to reflect the results, found in the first experiment, on.
Design public good experiments
Thirty participants are randomly assigned to one of six groups such that each group has five participants. Subsequently, three of these groups participated in the optional public good experiment and three participated in the base line experiment. The optional public good experiment is based on, and therefore quite similar, to the partner experiment of (Andreoni (1988). Hence, the experiment will be explained in comparison to his experiment.
The optional public good experiment is similar to Andreoni’s (1988) partner experiment in the following aspects. It has a duration of ten rounds, there are five participants per group and the MPCR is 0,5. The differences are the lack of a financial incentive for the participants, the use of an optional public good and a discrete per round endowment. In table 2 the similarities and differences between this and Andreoni’s (1988) experiment are orderly displayed.
Variables
Andreoni (1988)
Optional public good
experiment
Rounds 10 10
Group size 5 5
MPCR 0,5 0,5
Tokens/points 50 1
Public good type Non-optional Optional Financial compensation Sum of the per round earnings None
Table 2. Comparison of the variables used in Andreoni's (1988) and this experiment
Unfortunately, there were no funds available to compensate the participants accordingly to the decisions that they and the others in their groups made. Hence, the participants were asked to imagine as if the points they earned each rounds were actually euros. Camerer & Hogarth (1999) found that the missing of a financial incentive causes participants to make more virtuous decisions. This should be taken into account when the results are analysed.
A major difference is the use of an optional public good instead of a non-optional one. In standard public good experiments the participants are not given the choice whether they want to benefit from the public good. However, with optional public goods, it are only the users that benefit from the good. Hence, the participants are given the choice whether to use the public good or not. This choice has to be made simultaneously with the investment decision.
Furthermore, the participants in Andreoni’s experiment received fifty tokens to invest in the public and private good in any given division. The participants in the optional public good experiment received solely one point to invest in respectively the public or the private good. This was deemed necessary for several reasons. Firstly, the incentive to be a non-user would be pressured due to the fact that it more likely that a participant is investing an amount over zero to the public good. This reasoning comes from the assumption that many participants are likely to be conditional cooperators (Fischbacher et al., 2001; Fischbacher & Gächter, 2010; Keser & Van Winden, 2000) and therefore want to contribute at least something to the public good. Since they have to decide to be a user or a non-user of the public good simultaneously, the incentive to be a user is larger.
Secondly, the participant’s focus for this experiment should rather be on the use of the public good than on the proportional public good contribution. Determining what portion of the endowment should be spend on either of the two goods might become the main point of attention. By giving the participants a discrete choice, more emphasis is put on the public good use decision while the result of their decisions is also more insightful.
Lastly, the experiments had to be performed by hand since the knowledge and time to do the complete experiment using a computer program was missing. Calculating the per round earnings would have taken more time and the chance of errors would have been greater. Therefore, providing the participants with a single point was deemed the most effective solution.
Because of these two changes, the emphasis of the experiment has shifted. Whereas the experiment of Andreoni (1988) mainly focuses on the personal contribution towards the public good, this optional public good experiment points out for what reasons participants choose to invest in the private good over the public good. Especially the loner strategy is of particular interest to study this for several reasons. Firstly, the loner strategy is unavailable in standard public good games and therefore strictly associated with optional public good experiments. Furthermore, the loner strategy will only be chosen to minimize losses or punish free riders as it is weakly dominated by the free rider strategy when profit maximizing. The loner and free rider strategies solely yield the same profits when none of the participants chooses the contributor strategy.
The base line public good experiment is mostly the same as the optional public good experiment with the change that the public good is not optional. Hence, all participants benefit from a contribution to the public good. Therefore, the base line public good experiment is even more similar to Andreoni’s (1988) partner experiment. The differences are limited to the financial incentive and the endowment.
Implementation public good experiments
The groups were in turns invited to a common space at the Roeterseiland Campus of the University of Amsterdam. The participants were placed far apart in the room and were asked not to communicate in any way. Subsequently they were handed the instructions (Appendix A or C) and a fill in form (Appendix B or D) corresponding to their experiment.
On the fill in form of the optional public good experiment the participants had to fill in their investment and public good use decisions for each rounds. Furthermore, after each round, the supervisor filled in the total contribution to, as well as the number of users of, the public good. This enabled the participants to determine how many free riders, contributors and loners there were, assuming the number of altruists was zero. In the base line experiment, the participants solely had to fill in their investment decision as they by default make use of the public good. Besides, the supervisor filled in the total contribution to the public good after each round.
The participants were given approximately ten minutes to study both forms and were instructed to ask questions whenever something was unclear. Any questions raised were answered individually so the participants would not affect each other. After ten minutes, the supervisor of the experiment did a final check with everyone to make sure the instructions were understood. The participants were then put to work on the first round.
When all participants had made a decision for a round, for which they had approximately one minute, the fill in forms were collected by the supervisor. The supervisor then filled in the results as well as the individual earnings section of the fill in forms for that round. The supervisor sat at a distance from the participants to ensure that they would not be able to have a look at the forms of the others. Thereafter, the fill in forms were handed back to the participants. This process repeated itself until all ten rounds were played.
Results
Contributions to the public good in this experiment have to be interpreted slightly different than the contributions to a public good in other experiments (e.g. (Andreoni, 1988)). In this experiment the contribution to the optional public good is the outcome of the well-considered and discrete decision to invest in the private or the public good. Whereas, the contribution in previous public good experiments is the portion of the endowment that one is investing to the public good.
Base line experiment
The results of the base line experiment are displayed in figure 1. In the early rounds the average number of contributors is between the forty and sixty percent, similar to results found in other non-optional public good experiments (Ledyard, 1995). This level of cooperation is maintained until the sixth round before decreasing in the seventh, what is in accordance with results found by Andreoni (1988) and Croson (1996). However, in the last three rounds the contribution to the public good increases instead of decaying to the free rider level.
Figure 1. Strategies Base Line Public Good Experiment
Figure 1 is made out of the disaggregated data collected in three separate experiments. The decisions made in those separate base line experiments are displayed in figures 2, 3 and 4. In the first experiment there are more contributors than free riders for the first four rounds. Subsequently, the most chosen strategy starts to alternate for the remaining rounds. The second and third experiment are different from the first as there are more free riders than contributors in early rounds. However, the most picked strategy also starts to alternate eventually. This happens after the fifth or third round, for the second and third base line experiment respectively. Furthermore, none of the experiments show a lower level of contribution in the final rounds than at the start of the experiment. The second and third experiment even show a higher number of contributors in the final rounds, which explains the increase in cooperation towards the end of the experiment in the aggregated data graph.
0% 10% 20% 30% 40% 50% 60% 70% 80% 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Contributors Free riders
0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Contributors Free riders
0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Contributors Free riders
Optional public good
In this section there are two results that will discussed for the optional public good experiment. Firstly, the contribution to the public good will be analysed and compared to previous non-optional public good experiments. Secondly, the relative number of participants that choose the loner strategy is taken a look at.
In figure 5 the relative number of times a strategy is chosen by the participants is shown. In early rounds the relative number of contributors is between forty and sixty percent. This is comparable to the cooperation results found in standard public good experiments (Andreoni, 1988; Ledyard, 1995). The altruist and loner strategies are chosen by less than ten percent of the participants. The altruist strategy is most likely chosen because participants did not fully comprehend the payment structure yet. In later rounds this strategy is namely no longer used.
The loner strategy is, however, used trough out the experiment. This seems reasonable as the loner strategy is profitable, although less profitable than the free rider strategy. This strategy would therefore suit participants that do not choose to invest in the private good in order to maximize profit. These participants choose this strategy for one of the other reasons to invest in the public good. It could be because they expect none of the others to invest in the public good, causing them to have a lower profit would they choose the contributor strategy. Otherwise, it is possible that they choose this strategy to punish free ride behaviour by others. In both of these cases the participants apparently do not want to free ride themselves. Probably, these reasons are collectively responsible for participants to choose the loner strategy. Moreover, it is not possible to distinguish between these two strategies within this experiment. Hence, it can only be concluded that the loners are not interested in profit maximization.
Furthermore, in the final rounds, the participants will have a complete understanding of the experiment and the impact their decisions have on their earnings. Therefore, I would like to argue that the relative number of participants that choose the loner strategy in the final rounds is consciously making the decision not to free ride and therefore not to maximize profit. In the ninth round, the loner strategy is chosen over ten percent of the time and in the final round the strategy is chosen twenty percent of the time.
0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Contributors Free riders
Figure 5. Strategies Optional Public Good Experiment
Just as with the results of the base line public good experiment, the data shown in figure 5 is also derived from disaggregated data. The data of those experiments is displayed in figures 6, 7 and 8. In experiment 1 solely the contributor and free rider strategies were chosen. The loner and altruist strategies were not picked by any of the participants. Besides, the free rider strategy is chosen more often than the contributor strategy in all rounds of the experiment. In the second experiment there was more diversity in strategies chosen. The altruist strategy was still seldom picked, but the loner strategy was picked several times. Furthermore, the contributor and free rider strategies were chosen the most and alternated each other in terms of popularity. In the third experiment the loner strategy was chosen even more often than in the second experiment, while there were still rarely no altruists. Besides, for the first rounds there were more contributors than free riders. The tipping point was the fourth round from which onwards the number of free riders was higher or equal to the number of contributors.
All of the experiments show the same trend of decaying cooperation as the experiment progresses. In the early rounds the number of contributors is higher than in the final rounds. Moreover, for all of the experiments the number of free riders is higher in the final rounds than in the first. A remarkable observation is the lack of loners and altruists in the first experiment. In the other two experiments both strategies are chosen at least once. Although, as stated before the altruist strategy is only chosen in early rounds, probably due to a limited understanding of the payment structure. The loner strategy is chosen throughout in both of the experiments.
0% 10% 20% 30% 40% 50% 60% 70% 80% 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Comments
One of the possible explanations for the strange results in the base line experiment could be a difference in the earnings for contributors in both experiments. When the difference in earnings between free riders and contributors is small, the incentive to become a free rider is less present. Hence, the difference in earnings is supposed to be smaller in the base line experiment as there are more contributors in the final rounds than in the optional public good experiment.
In each public good experiment the absolute difference in earnings between contributors and free riders is the additional income earned through the private good. Therefore, the absolute difference in earnings is independent from the level of contribution. However, the relative difference in earnings is affected by the level of contribution. When there are few contributors, the relative difference between the earnings of free riders and contributors is high, whereas the relative difference is low when there are many contributors. Therefore, to determine whether a difference in earnings could explain the increase in cooperation in the final rounds, the relative earnings of the contributors in both experiments have to be compared to each other.
Figure 6. Strategies Optional Public Good Experiment 1 Figure 7. Strategies Optional Public Good Experiment 2
0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Contributors Free riders Loners Altruists 0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Contributors Free riders Loners Altruists 0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
STRATEGIES
Contributors Free riders Loners Altruists
Figure 9. Relative earnings of contributors compared to free riders
In figure 9 the relative earnings of the contributors in both experiments are displayed. In the first two rounds the relative earnings of contributors is higher in the optional public good experiment than in the base line experiment. In rounds three, four, six and eight the relative earnings are higher in the base line experiment, while it is equal for the fifth round. The seventh round is the final round in which the relative earnings are higher in the optional public good than in the base line experiment. So until the final rounds, the relative earnings are three times higher in the optional public good experiment and four times higher in the base line experiment. To conclude whether this is a significant difference, a Wilcoxon rank-sum test has been conducted. The null hypothesis was that the relative earnings of the base line experiment (type 0) were equal to the relative earnings of the optional public good experiment (type 1) for the first eight rounds. The results of the Wilcoxon rank-sum test are displayed in table 3 below. From the p-value (0.4538) it has to be concluded that the null hypothesis cannot be rejected. Hence, the relative earnings of the base line experiment are not significantly higher than the relative earnings of the optional public good experiment.
Thus, the increase in contribution for the final rounds of the base line experiment cannot be explained by a difference in earnings between the two experiments. Another explanation can be the lack of compensation for the participants. Participants tend to give more favourable answers when there is no financial incentive (Camerer & Hogarth, 1999). Although, this would suggest that the results in the
0% 10% 20% 30% 40% 50% 60% 70% 80% 1 2 3 4 5 6 7 8 9 1 0 ROUNDS
RELATIVE EARNINGS
Contributors optional public good experiment Contributors base line public good experiment
Prob > |z| = 0.4538 z = 0.749
Ho: REL_EA~S(Type==0) = REL_EA~S(Type==1) adjusted variance 87.33
adjustment for ties -3.33 unadjusted variance 90.67 combined 16 136 136 1 8 61 68 0 8 75 68 Type obs rank sum expected Two-sample Wilcoxon rank-sum (Mann-Whitney) test
Prob > |z| = 0.4538 z = 0.749
Ho: REL_EA~S(Type==0) = REL_EA~S(Type==1) adjusted variance 87.33
adjustment for ties -3.33 unadjusted variance 90.67 combined 16 136 136 1 8 61 68 0 8 75 68 Type obs rank sum expected Two-sample Wilcoxon rank-sum (Mann-Whitney) test
optional public good experiment should also have been affected, what does not seem to be the case. Therefore, other explanations have to be assessed. Two other explanations are the uncertainty surrounding the motive to free ride in the base line experiment and the complexity of the optional public good experiment.
In the base line public good experiment, participants are unable to ensure a profit of at least one while not free riding. The motivation to invest in the private good over the public good could be the result of any of the before mentioned reasons for an investment in the private good. Hence, the incentive of other participants to invest in the private good in the base line public good experiment cannot be determined. This most likely leads to participants being more forgiving to free riders in the base line experiment than in the optional public good experiment which results in more contributors in the final rounds.
Furthermore, the complexity of the optional public good experiment might have been an important factor in the lower level of cooperation. Due to the complexity, the participants are prevented from easily understanding what the virtuous decisions are. Therefore, the focus of these participants is on deciding what strategy fits their incentives best, whereas the focus of the participants in the base line experiment might have shifted to presenting themselves as social human beings.
Conclusion
The level of cooperation in the optional public good experiment is similar to the results found in non-optional public good experiments studies. Furthermore, the number of free riders in the early rounds is in accordance with previous studies. However, the number of free riders in the final rounds is lower than in those studies due to participants that choose the loner strategy over the free rider strategy. This implies that in non-optional public good experiments some participants invest in the private good for another reason than pure profit maximizing. This is in accordance with previous studies and would suggest that in standard public good experiments some participants are free riding because they want to minimize losses or punish free riders.
On the contrary, the level of cooperation in the base line experiment is not similar to previous public good experiments. For the first eight rounds, the results match those of other studies. However, in the final rounds the number of contributors increases which is very unlike previous studies and cannot be explained by economic theory. Two explanations are the uncertainty related to the motive of other participants to free ride and the complexity of the public good.
Discussion
The participants in both experiments are acquaintances of the author and most knew one or more of the other participants in their group. This might have affected their decisions since it was not clarified in the instructions whether their answers would be available to others when the experiment was done. Moreover, the number of participants and groups might have been slightly too minimal. In other public good experiment there are usually at least forty subjects.
Furthermore, in this experiment the payment structure of a public good is used. Thus, the use of the public good by any of the participants will not affect the possible earnings of others. Therefore, the impact of free riding is solely morally. A common good might show kindness and punishment behaviour even better as free riders willingly decrease the earnings from the public good for contributors.
References
Andreoni, J. (1988). Why free ride?: Strategies and learning in public goods experiments. Journal of
Public Economics, 37(3), 291-304.
Andreoni, J. (1995). Cooperation in public-goods experiments: Kindness or confusion? The American
Economic Review, , 891-904.
Bohm, P. (1972). Estimating demand for public goods: An experiment. European Economic Review,
3(2), 111-130.
Camerer, C. F., & Hogarth, R. M. (1999). The effects of financial incentives in experiments: A review and capital-labor-production framework. Journal of Risk and Uncertainty, 19(1-3), 7-42.
Chaudhuri, A. (2011). Sustaining cooperation in laboratory public goods experiments: A selective survey of the literature. Experimental Economics, 14(1), 47-83.
Croson, R. T. (1996). Partners and strangers revisited. Economics Letters, 53(1), 25-32.
Fehr, E., & Gächter, S. (1999). Cooperation and punishment in public goods experiments. Institute for
Empirical Research in Economics Working Paper, (10)
Fischbacher, U., Gächter, S., & Fehr, E. (2001). Are people conditionally cooperative? Evidence from a public goods experiment. Economics Letters, 71(3), 397-404.
Fischbacher, U., & Gächter, S. (2010). Social preferences, beliefs, and the dynamics of free riding in public goods experiments. The American Economic Review, 100(1), 541-556.
Gneezy, U., & Rustichini, A. (2000). Fine is a price, a. J. Legal Stud., 29, 1.
Isaac, R. M., & Walker, J. M. (1988). Group size effects in public goods provision: The voluntary contributions mechanism. The Quarterly Journal of Economics, , 179-199.
Keser, C., & Van Winden, F. (2000). Conditional cooperation and voluntary contributions to public goods. The Scandinavian Journal of Economics, 102(1), 23-39.
Kim, O., & Walker, M. (1984). The free rider problem: Experimental evidence. Public Choice, 43(1), 3-24.
Ledyard, J. (1995). Public Goods: Some Experimental Results, Chapter 2 in John Kagel and Alvin Roth (eds.) Handbook of Experimental Economics
Olson, M. (1965). The logic of collective action: Public goods and the theory of groups. Cambridge,
Mass,
Sweeney, J. W. (1973). An experimental investigation of the free-rider problem. Social Science
Research, 2(3), 277-292.
Weimann, J. (1994). Individual behaviour in a free riding experiment. Journal of Public Economics,
54(2), 185-200.
Zelmer, J. (2003). Linear public goods experiments: A meta-analysis. Experimental Economics, 6(3), 299-310.
Appendix A – Instructions optional public good experiment
Introduction
In this experiment you can earn points based on the decisions that you and the other participants in your group make. It is very important that you act as if every point was a euro. After the final round your total earnings will be accredited to you.
Group
At the start of the experiment, you will form a group of 5 with 4 other participants. During the entire experiment you will stay in this group.
Rules
The experiment consists of 10 rounds. At the start of each round you will receive an endowment of 1
point. Subsequently, there are two decisions you have to make:
1) Investment decision: the point you receive each round has to be invested in one of two goods. You can invest it in a private good or in a public good. Every investment in the private good benefits solely yourself, while all users of the public good benefit from an investment in the public good.
2) Public good use decision: for every round you have to decide whether you want to make use of the public good or not. Therefore, you can be a user or a non-user of the public good. Users receive points from the private good and the public good, while non-users only receive points from the private good.
The decisions you make are not visible to the other participants. However, at the end of each round the amount of points contributed to the public good and the total number of users is shown.
Payment
Your per round earnings are the sum of the returns from the private and the public good. The formula for the per round earnings looks as follows:
𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 = 𝑃𝑟𝑖𝑣𝑎𝑡𝑒 𝐺𝑜𝑜𝑑 + 𝑃𝑢𝑏𝑙𝑖𝑐 𝐺𝑜𝑜𝑑
Private good
The return from the private good is equal to the investment in the private good. So for every point you invest in the private good, you will receive one point. Therefore the earnings from the private good can be shown as:
𝑃𝑟𝑖𝑣𝑎𝑡𝑒 𝐺𝑜𝑜𝑑 = 𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 𝑖𝑛 𝑃𝑟𝑖𝑣𝑎𝑡𝑒 𝐺𝑜𝑜𝑑
Public good
Different from the private good, the earnings from the public good are based on the collective contribution to the good. The earnings from the public good are equal to a half times the total points. So for every point invested in the public good, all users receive half a point. Hence, the earnings from the public good can be shown with the following equations:
{
𝑃𝑢𝑏𝑙𝑖𝑐 𝐺𝑜𝑜𝑑 =1
2∙ 𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑖𝑛𝑡𝑠 if user 𝑃𝑢𝑏𝑙𝑖𝑐 𝐺𝑜𝑜𝑑 = 0 if non-user
Earnings
Since you have two choices for both the investment and the public good use decision, there are four different strategies. The earnings from these strategies are summarized in the table below.
Investment
Private good
Public good
Public good use
Yes
𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 = 1 + ½ ∙ 𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑖𝑛𝑡𝑠 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 = 0 + 1 2∙ 𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑖𝑛𝑡𝑠No
𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 = 1 + 0 = 1 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 = 0 + 0 = 0 Fill in formOn the fill in form you can indicate your decisions regarding your investments and the public good use for each round. You can do so by marking your choice with a cross in the designated cell. It is important that you fill in your decisions for one round at a time only. After you made your decisions for a round, hand in the form. The results of that round will then be analysed and shared with you.
Appendix B – Fill in form optional public good experiment
Appendix C – Instructions base line public good experiment
Introduction
In this experiment you can earn points based on the decisions that you and the other participants in your group make. It is very important that you act as if every point was a euro. After the final round your total earnings will be accredited to you.
Group
At the start of the experiment, you will form a group of 5 with 4 other participants. During the entire experiment you will stay in this group.
Rules
The experiment consists of 10 rounds. At the start of each round you will receive an endowment of 1
point. Subsequently, you can invest the point in one of two goods. You can invest it in a private good
or in a public good. Every investment in the private good benefits solely yourself, while everyone benefits from an investment in the public good. The decisions you make are not visible to the other participants. However, at the end of each round the amount of points contributed to the public good is shown.
Payment
Your per round earnings are the sum of the returns from the private and the public good. The formula for the per round earnings looks as follows:
𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 = 𝑃𝑟𝑖𝑣𝑎𝑡𝑒 𝐺𝑜𝑜𝑑 + 𝑃𝑢𝑏𝑙𝑖𝑐 𝐺𝑜𝑜𝑑
Private good
The return from the private good is equal to the investment in the private good. So for every point you invest in the private good, you will receive one point. Therefore the earnings from the private good can be shown as:
𝑃𝑟𝑖𝑣𝑎𝑡𝑒 𝐺𝑜𝑜𝑑 = 𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 𝑖𝑛 𝑃𝑟𝑖𝑣𝑎𝑡𝑒 𝐺𝑜𝑜𝑑
Public good
Different from the private good, the earnings from the public good are based on the collective contribution to the good. The earnings from the public good are equal to a half times the total points. So for every point invested in the public good, everyone receives half a point. Hence, the earnings from the public good can be shown with the following equation:
𝑃𝑢𝑏𝑙𝑖𝑐 𝐺𝑜𝑜𝑑 =1
2∙ 𝑡𝑜𝑡𝑎𝑙 𝑝𝑜𝑖𝑛𝑡𝑠
Fill in form
On the fill in form you can indicate your decisions regarding your investments for each round. You can do so by marking your choice with in cross in the designated cell. It is important that you only fill in your decision for one round at a time. After you made your decision for a round, hand in the form. The results of that round will then be analysed and shared with you.
