EVIDENCE-BASED RESEARCH ON CHARITABLE GIVING SPI Funded Voluntary and Mandatory Provision of Common Pool Resources with Heterogeneous Appropriators Fatemeh Momeni Purdue University SPI Working Paper No.: 131 - SPI January 2015 Voluntary and Mandatory Provision of Common Pool Resources with Heterogeneous Appropriators∗†‡ Fatemeh Momeni Purdue University January 2015 Abstract Common pool resource (CPR) users often face two types of problems to solve: provision problems and appropriation problems. This paper presents a laboratory experiment to study the choices of CPR users under different provision schemes, in a heterogeneous environment. In the first two treatment conditions, the level of contribution to the provision process is determined exogenously: in the regressive treatment, poor and rich individuals pay equal amounts to the provision of the resource, and in the progressive treatment, the entire provision cost is paid by rich members. Finally, in the endogenous condition subjects voluntarily choose how much to contribute through the Provision Point Mechanism. The experimental results provide strong evidence for inequality aversion motivating subjects’ decisions. Interestingly, inequality aversion motivates the subjects’ choices differently in the exogenous and in the endogenous provision conditions: in the regressive and the progressive treatments inequality aversion motivates the extraction choices while in the endogenous treatment inequality aversion motivates subjects’ contribution decisions but not their extraction decisions. ∗ This research is funded by Science of Philanthropy Initiative and John Templeton Foundation. I am grateful to Timothy Cason for his continuous guidance and encouragement. I would like to thank Sandra Maximiano, Steven Wu and Mohitosh Kejriwal for valuable comments and discussions. All remaining errors are my own. ‡ Contact: Department of Economics, Purdue University, West Lafayette, IN 47907, USA. Tel: +1 (765) 409-1286. E-mail: [email protected] The latest version of this paper is available at: https://sites. google.com/site/fatemehmomeni/research † 1 1 Introduction Ostrom, Gardner and Walker (1994) classified problems facing Common Pool Resource (CPR) users into two types: provision problems, which deal with creating the resource, maintaining it, and improving its production capabilities; and appropriation problems, which relate to allocating the subtractable flow from the resource among potential users. It is natural to assume that these two classes of problems are linked to one another. Ostrom, Gardner and Walker (1994) points out that “the nature of appropriation problem is affected by how well the provision problem is solved”. The link between provision and appropriation problems could be explained by the idea that people derive procedural utility (Frey, Benz and Stutzer (2004)); suggesting that individuals care not only about the outcomes, but also about how those outcomes are achieved. In the case of CPR, the idea of procedural utility implies that, in addition to the utility that resource users receive from appropriating the resource, they may derive utility – or disutility – from the way in which the resource is provided to them. Thus, understanding the interaction between provision and appropriation problems serves as an essential move towards understanding CPR users’ extraction decisions in facing such social dilemmas.1 Moreover, recent studies have argued that by accounting for the fact that CPR users often interact with each other in other social dilemma situations, we can explain some real world observations that are not replicated in the laboratory setting which study the appropriation problems in isolation.2 One social dilemma situation through which CPR users interact with each other concerns solving the provision problem: all users are better off if the resource is provided, and yet each individual has an incentive to not contribute in the provision process. Therefore, by understanding the interaction between provision and appropriation problems, we may overcome some shortcomings of previous CPR experiments in creating a laboratory environment that better captures the properties of CPR in the field settings. Despite recognizing the importance of studying provision and appropriation problems jointly, nearly all previous CPR experiments have taken the provision of CPR as given, and studied the appropriation problem in isolation. Many CPRs are provided or maintained through the contributions of their users who directly or indirectly contribute from their private resources in the provision process. Main1 Previous studies have found evidence that origin of resources affects how people use those resources (Thaler, 2008; Arkes et al., 1995). 2 See Botelho et al. (2013) for an experimental and Solstad and Brekke (2011) for a theoretical study. 2 tenance, repair and routine operational activities by irrigation system users are examples of direct contribution of users in the provision process. In some other CPRs, such as government-run fisheries, national parks and forestries, users indirectly contribute in provision activities, by paying taxes from their wealth to a central authority who runs, manages and maintains the common resource. CPR users are often heterogeneous in various dimensions such as their wealth, their benefit from appropriating the resource, their ability to extract from the resource, their outside option, etc. The presence of heterogeneity may create multiple and conflicting viewpoints about how the provision and appropriation problems should be solved: should poor users pay the same share of the provision costs as rich users do? Should those who benefit more from extraction, contribute more in the provision process? Should farmers with more land be allowed to extract more water from irrigation canals? Should fishers endowed with more fishing equipment be allowed to harvest more from common fisheries? Viewing the procedures to be fair has been shown to increases people’s propensity to cooperate in social dilemmas [Trawick (2001)]. Since monitoring appropriation activities in CPR situations if often very costly, finding a provision scheme that appears fair and appeals to most users may increase cooperation in conserving and better utilizing the resource, without allocating a lot of resources into costly monitoring procedures. In this paper I design a laboratory experiment to explore how different provision mechanisms affect CPR users’ extraction choices in a heterogeneous environment. The heterogeneity is introduced in subjects’ wealth. Rich and poor individuals – with different abilities to contribute in the provision process – participate in 20 rounds of a provision-appropriation game. I compare individual decisions under three provision schemes. In the first two treatments, the level of contribution in the provision process is determined exogenously: in the regressive treatment, all users pay the same share of the cost of providing the resource, while in the progressive treatment, the entire provision cost is paid by rich individuals.3 4 In the third – endogenous – treatment, users voluntarily contribute from their wealth to the provision process. Previous studies have found that the endogenous choice of the process increases 3 Equal payment by rich and poor is a special case of regressive tax in the sense that poor individuals pay a higher portion of their wealth in taxes. There are many examples of regressive provision of common resources practiced in the field. In the US for instance, many states issue commercial and recreational fishing permits with fixed fees (See for instance Alaska Department of Fish and Game at http://www.adfg.alaska.gov/, or Florida Fish and Wildlife Conservation Commission at http://myfwc.com/). Other examples are national parks and public facilities that charge their users fixed admission fees. 4 In the progressive treatment, rich players pay a higher portion of their wealth in the provision process. Such practices are also common in the field. For example Trawick (2001) studies water management in Andean communities in Peru where larger land owners are required to contribute more than small land owners in provision activities such as repair and maintenance. 3 the level of cooperation.5 By comparing the extraction choices under the endogenous and the exogenous provision schemes, I explore whether voluntary provision of the resource can lead to higher level of cooperation in extracting from it. Moreover, studying users’ contribution choices in the endogenous provision scheme can shed light on their preferences about how the provision costs of the resource should be divided among its users. Under the standard assumption of expected money maximization, the way in which resource is provided and the existing heterogeneity in the provision stage does not impact users’ extraction choices. However, two behavioral models of inequality aversion and Equity theory predict different extraction patterns for rich and poor individuals and under different provision schemes. Equity theory [Adams (1963), Homans (1961), Konow (2003)] suggests that those who contributed more in the provision of the resource should extract more from the resource, once the resource is provided for use. Inequality aversion theory (Fehr and Schmidt 1999) predicts that rich individuals contribute more in the provision of the resource (in the endogenous treatment), and extract less from the resource to reduce inequality. The experimental results clearly refute the predictions of the Equity theory and provide strong evidence for inequality aversion motivating subjects’ decisions. Interestingly, inequality aversion motivates subjects’ choices differently in the exogenous and in the endogenous provision conditions: in the regressive and the progressive treatments, in which the level of contribution in the provision process is not subjects’ choice, inequality aversion motivates the extraction choices – rich subjects extract significantly less than poor subjects do. By contrast, in the endogenous treatment, inequality aversion motivates subjects’ contribution decisions – rich subjects contribute significantly more than poor subjects do – but not their extraction decisions. The fact that inequality aversion motivates the extraction choices in the exogenous treatments but not in the endogenous treatment suggests that the voluntary nature of subjects’ previous actions in the provision stage to reduce the inequality in the endogenous treatment has reduced their degree of discomfort from the inequality. This observation is consistent with the findings of previous experiments about the role of intentionality in explaining individual choices (Blount 1995; Offerman 2002; Charness 2004). I do not find a significant difference in aggregate extractions across the three treatment conditions. At the same time, among the three provision schemes, the level of inequality is reduced the most under the the progressive provision condition. Thus, if the goal is to reduce 5 Wahl, Muehlbacher and Kirchler (2010); Feld and Frey (2002); Sutter, Haigner and Kocher (2010) 4 the inequality and to preserve the resource at the same time, the progressive provision rule proves to be the most successful provision scheme among the three that are studied here. Finally, studying individual responses to the choices made by others reveals an asymmetry in subjects’ responses to their group members’ choices, depending on their type: subjects increase their extraction as the same-type players in their group extract more from the resource, and decrease their extraction as the opposite-type players extract more from the resource. This is the first experimental study that compares CPR users’ choices under voluntary and mandatory provision schemes. It is also the first study that compares different provision mechanisms in a heterogeneous environment. I am aware of three other studies that investigate the interaction between provision and appropriation problems using laboratory experiments. The experimental design of the first two papers (Cardenas, Rodriquez and Johnson (2011) and Janssen, Anderies and Joshi (2011)) are motivated by problems facing irrigation system users who have asymmetric access to the resource. In both experiments, subjects first decide how much to contribute to the provision of the resource. The aggregate contribution level determines the flow of the resource that becomes available for use. Next, subjects sequentially decide how much to extract from the available flow. Only the flow that is not extracted by previous (upstream) users is available for use to the next (downstream) users. Perhaps the closest experimental design to this paper is Cherry, Cotton and Jones (2013). The authors study the extraction choices of CPR users who first endogenously decide whether to voluntarily contribute some of their endowments towards the enhancement of the resource, and then decide how to allocate their remaining endowments between extracting from CPR and a private alternative with a fixed return. While the first two papers introduce asymmetry in the appropriation stage, Cherry, Cotton and Jones (2013) is completely symmetric in both stages. In contrast with these three papers, the design of the current paper introduces heterogeneity in the provision stage and is completely symmetric in appropriation stage. An important feature of the experimental environment in the current paper is heterogeneity. Here the heterogeneity concerns differences in the ability to contribute in the provision of the resource whereas some other forms of heterogeneity in CPR environments have been explored in previous experimental studies. Hackett, Schlager and Walker (1994) studies the effect of communication in increasing cooperation among heterogeneous CPR users, who 5 have asymmetric endowments of resources to use in extraction. Margreiter, Sutter and Dittrich (2005) compares coordination on efficient use of a CPR using a voting scheme, between groups with homogeneous and heterogeneous cost of appropriation. Cardenas, Strandlund and Willis (2002) designs a CPR experiment in which users are heterogeneous in their return from their outside options. Finally, Hayo and Vollan (2012) studies the effect of heterogeneity in real-world income, on extraction behavior of CPR users in a laboratory setting. The remainder of this paper is organized as follows: section 2, explains the experimental design and discusses the treatments and the experimental parameters. Section 3, discusses the equilibrium solutions for purely self-interested player and explores how inequality aversion and Equity theories change these predictions. Sections 4 and 5, present and discuss the experimental results and section 6 concludes. 2 Experimental Design, Treatments and Parameters I design a CPR provision-appropriation experiment to explore how different provision schemes affect users’ extraction choices. I compare the extraction choices under three provision mechanisms: regressive, progressive and endogenous. Users’ contribution to the provision process is determined exogenously in the regressive and the progressive treatments, and endogenously in the endogenous treatment. The experiment is composed of two sequential stages: the provision stage followed by the appropriation stage. Subjects enter the provision stage with unequal wealth, and according to the treatment condition – either voluntarily or mandatorily – contribute from their wealth to the provision process. After the resource is provided, subjects participate in the completely symmetric appropriation stage in which they decide how to use the common resource. To help subjects understand the experimental instructions, subjects’ choices are framed as decisions about the use of a fishery in a given year.6 Below, I explain the design of the two stages of the game and the parameter choices for each treatment condition. 6 Framed experimental instructions has been previously used in some CPR experiments to help subjects better understand the characteristics of the game. e.g. Akpalu and Martinsson (2009); Cardenas, Rodrigues and Johnson (2011) ; Janssen, Anderies and Joshi (2011); Cardenas and Ostrom (2004) 6 Provision Stage: Users enter the provision stage with unequal wealth. Groups are composed of two rich and two poor individuals (n=4) endowed with 160 and 20 experimental dollars at the beginning of the provision stage. The cost of providing the resource is 67.2 experimental dollars which is collected from the users according to the treatment condition. The experimental instructions explain to the subjects that the collected fund is used to run hatcheries essential for provision and preservation of the fishery. In the regressive treatment, all individuals pay the same share of the provision cost (16.8 experimental dollars) regardless of their wealth level, and in the progressive treatment, the entire provision cost is paid by the rich members (33.6 experimental dollars each). In the third – endogenous – treatment, users endogenously decide how much to contribute to the provision of the resource through Provision Point Mechanism (PPM). The provision threshold is kept at the same level of 67.2 experimental dollars and all four individuals, simultaneously and independently choose their contribution levels. If the total contribution by the group reaches the threshold, the resource is provided for use. To encourage contribution and increase the likelihood of successful provision, I use the refund rule, meaning that the contributions are returned to the contributors if the provision is not successful.7 In the case of unsuccessful provision, the experimental year ends after the provision stage, and the users only earn their initial wealth. Some previous studies have established that endogenous choosing of the economic setting can lead to higher individual cooperation (e.g. Wahl, Muehlbacher and Kirchler (2010); Feld and Frey (2002), Sutter, Haigner and Kocher (2010), Frey et al (2004)). By comparing the extraction choices in the endogenous treatment to those in the regressive and progressive treatments, I explore whether the endogenous choice of the contribution level in the provision stage can lead to more cooperative choices in extraction from the common resource. Appropriation Stage: Following the provision stage, all groups in the regressive and the progressive treatments, and those in the endogenous treatment who successfully provided the resource enter the appropriation stage. Unlike the provision stage, the appropriation stage is completely symmetric. The design of the appropriation stage closely follows the baseline appropriation experiment first introduced by Ostrom, Gardner and Walker (1994), 7 However, in the case of success, the excess contributions are not returned. Thus, groups have to coordinate on contributing no more than the required threshold to reach a more efficient outcome. For more information about different rebate rules in threshold public good experiment see Marks and Croson (1998). 7 which has been widely used in previous CPR experiments: N individuals with access to the CPR are each endowed with e = 12 units of endowment of private resource. Each unit of e can be spent either in extracting from the common resource, or in another private activity with a fixed return of w = 2.8 The payoff from the appropriation stage to an individual who uses xi units of e in extracting from the CPR is given by: Mi (xi ) = w(e − xi ) + X X xi [a x − b( xi )2 ]. i N P xi (1) i=1 The first and the second terms represent the return from the part of endowment e that the agent spends in the private activity and in extracting from the common resource. The payoff from extraction depends not only on individual i’s extraction (xi ), but also on the P aggregate group extraction , xi . The quadratic term in the brackets corresponds to CPR P production function, F ( xi ). The parameters must be chosen such that for small values of group extractions, spending the endowments in extracting from the common resource pays better than the private activity (F 0 (0) = a > w), but if the group extracts too much from the resource, the outcome is counter-productive (F 0 (N e) = a − 2bN e < 0). The values for the parameters a(= 18), b(= 0.4), e(= 12) and w(= 2) are taken from Cason and Gangadharan (2014). The asymmetric-provision, symmetric-appropriation design of the experiment leaves the action space identical for all individuals in the appropriation stage. The symmetric design of the appropriation stage leads to the same equilibrium prediction for extraction choices by purely self-interested rich and poor players. Thus, any observed differences in extraction choices by rich and poor must have been caused by behavioral motives resulting from adding the provision stage to the game. Information structure: Before entering the appropriation stage, subjects fully observe the level of contribution in the provision process by all other group members.9 The information 8 In experimental instructions, each unit of private endowment e represents one month in the current experimental year. Subjects can spend each of the months either in a private activity, or in fishing from the common fishery. 9 Due to the mandatory nature of contributions in the regressive and progressive treatments, this information is already known by all individuals in these two conditions. However, to keep the information structure in the three treatments comparable, this information is again reminded to all subjects, before entering the 8 about individual contributions in the provision process is given to the subjects to provide the grounds for potential behavioral-type players to judge the fairness of the allocations, when making their extraction choices. Moreover, at the end of each round, subjects are informed about each individual’s extraction choices and payoffs. Receiving information about contribution and extraction choices at the individual level regularly happens in the field (Ostrom, Gardner and Walker 1994). It should be noted that one previous study (Villena and Zechhetto 2011) finds that providing individual-specific information about subjects’ choices makes subjects less cooperative in solving the appropriation problem. We may thus expect to observe a similar effect in this experiment as well.10 The provision-appropriation game is repeated for 20 rounds and subjects remain in the same groups (partner-matching) for all the 20 rounds of the experiment. The duration of the experiment is public information. The earning activity: To give subjects a sense of entitlement to their experimental wealth, I add an earning stage which takes place before the start of the 20-round experiment. Previous studies have shown that earned income, as opposed to windfall income, increases individuals’ sense of entitlement to their wealth, which plays important roles in everyday economic decision makings.11 Prior to receiving the instructions for the 20-round experiment, subjects take a 10-question multiple-choice quiz. Quiz questions are chosen from sample GMAT multiple-choice questions. Subjects are informed that their performance in the quiz determines the amount of experimental wealth they are going to receive at the beginning of each round of the experiment that follows. Specifically, subjects are instructed that half of the participants with higher scores in the quiz receive 160 experimental dollars and the rest receive 20 experimental dollars, at the beginning of each of the 20 experimental rounds.12 appropriation stage. 10 Villena and Zechhetto (2011) study CPR users’ extraction choices under different information structures. They find that individuals make less cooperative choices, when given individual-specific information about other players’ choices, and conclude that individual-specific information worsens the tragedy of commons. 11 For instance individuals make more generous and more risky choices with windfall money. [Hoffman et al (1994)] 12 This information is given to motivate subjects to put effort in solving quiz problems. Additionally, subjects learn the exchange rate between experimental and real dollars before taking the quiz. 9 3 Theoretical Predictions To make predictions about the extraction and the contribution choices in different treatment conditions, this section stats by finding the equilibrium choices by purely self-interested, own-payoff maximizing players. Next, it explores how these equilibrium predictions are modified if inequality aversion and Equity theory motivate individual choices, and discusses how the solutions under these two theories differ under each treatment condition. 3.1 Regressive and Progressive Provision Schemes In order to find the equilibrium predictions for extraction levels by purely self-interested individuals, let’s start by considering the appropriation stage. This second stage of the game is completely symmetric and identical across all treatments. The monetary payoff to an individual who spends xi units of her endowments in extracting from the common resource is given by: we Mi (xi ) = w(e − xi ) + xi (a N P xi P xi − b( P xi = 0 xi )2 ) xi > 0. i=1 Taking the first derivative from this payoff function and imposing symmetry, it is easy to verify that the equilibrium level of extraction is achieved when each player spends x∗i = 8 units of her endowment e in extracting from the resource, 13 while the social optimum level P so of extraction is reached when the group as a whole, spends xi = 20 units in extraction. P so P ∗ Note that xi < xi (= 8 × 4 = 32). Adding the provision stage does not change the equilibrium prediction of extraction for an own-payoff maximizing player. For a self-interested subject, the way in which the resource is provided is irrelevant to her extraction decisions. Let’s now explore how the two behavioral theories of inequality aversion and Equity Theory change the predictions from that of the standard self-interested model. Not only x∗i is the symmetric equilibrium level of extraction, but the parameters are chosen such that = 8 is the only equilibrium solution for the own-payoff maximizing player in this environment. 13 x∗i 10 Equity theory [Adams (1963), Homans (1961), Konow (2003)] suggests that the just allocation is achieved, when each individual’s reward is proportional to her input (∀i, j : Input Inputi = Rewardj j ). If subjects view their contributions to the provision of the resource as Rewardi the right input variable, they may use Equity Theory to judge the fairness of the allocations. In the regressive treatment, in which all four players contributed the same amount towards the provision of the resource, Equity Theory predicts the same level of extractions by rich and poor (xrich = xpoor ). In the progressive treatment however, in which rich players paid the entire provision costs, Equity theory predicts a higher extraction by rich individuals (xrich > xpoor ; xpoor = 0). The experimental design imposes inequality in the provision stage. The parameters are chosen such that this inequality can never be undone regardless of the players’ extraction choices. So one may conjecture that – for inequality-averse enough players – inequality aversion motivates the extraction choices. Given that subjects receive full information about the choices and payoffs of all individuals in their group –which makes payoff comparisons at individual level feasible– I use Fehr and Schmidt (1999) inequality aversion model to predict the extraction choices. Fehr and Schmidt (1999) formulates the utility of an inequality P P averse individual as Ui = πi − αi N1−1 j6=i max{πj − πi , 0} − βi N1−1 j6=i max{πi − πj , 0}14 , in which πk represents the payoff to individual k. The second and the third terms represent the utility losses from disadvantageous and advantageous inequality, respectively. A rich player with high enough βi (advantageous inequality aversion) may increase her utility by reducing her extraction from the CPR and reducing the initial inequality. Since the initial inequality between rich and poor is relatively large in both the regressive and the progressive treatments, inequality aversion theory predicts higher extraction by poor players under both treatment conditions, for inequality averse enough individuals. Table 1 summarizes the above predictions. 3.2 Endogenous Provision Scheme Subjects in the endogenous treatment condition make two sequential decisions. First, they decide how much to contribute from their unequal wealth to the provision of the resource 14 Here, πi = Wi − Ti + Mi (xi ), in which Wi is individual wealth received at the beginning of the provision stage, Ti represents individual contribution to the provision process, and Mi (xi ) represents individual payoff from the appropriation stage. 11 Table 1: Predictions of Inequality Aversion Theory and Equity Theory for the Regressive and the Progressive Treatments Regressive Trich = Tpoor = 16.8 Equity Theory xrich = xpoor Inequality Aversion (For high enough α and β) xrich < xpoor Progressive Trich = 33.6; Tpoor = 0 xrich > xpoor ; xpoor = 0 xrich < xpoor (choose Ti ). Next – and if the group successfully provides the resource – individuals decide how to allocate their endowment of 12 tokens (months) between extracting from the CPR and the private activity with constant return (choose xi ). To find the equilibrium decisions of this provision-appropriation game for a purely self-interested individual, I use backwards induction. Similar to that of the regressive and the progressive treatments, the equilibrium level of extraction in the second – appropriation – stage is achieved when each player spends x∗i = 8 units of her endowment in extraction. Playing this equilibrium in the appropriation stage leads to individual payoff of Mi (x∗i ) = 49.6 experimental dollars from the second stage. As long as individual contribution does not exceed this payoff, it is worthwhile to contribute towards providing the resource. However, since the required provision threshold (67.2) is larger than Mi (x∗i ), a self-interested player will not pay the entire provision costs of 67.2 by herself.15 The provision stage has one trivial equilibrium in which no-one contributes any amount, and many other non-trivial equilibria that satisfy three conditions: 1) the resource is proP vided; 2) the contributions are efficient ( Ti = 67.2) and 3) no individual contribution exceeds Mi (x∗i ) = 49.6 (∀i : Ti < Mi (x∗i )). Input Inputi = Rewardj j ), we should If individual decisions are motivated by Equity theory ( Reward i observe those individuals who contributed more in the provision stage to extract more from T the resource in the appropriation stage ( Txii = xjj ). However, if users are enough inequality averse, they may increase their utility by making contribution and extraction choices that result in a reduction in inequality. This suggests that rich individuals with high enough advantageous inequality aversion (βi ) extract less in the appropriation stage (xrich < xpoor ) or contribute more in the provision stage (Trich > Tpoor ), while poor players with high enough disadvantageous inequality aversion (αi ) extract more from the resource in the appropriation 15 Note that paying the entire provision cost is only feasible for rich players. Poor players’ wealth (Wpoor = 20) is less than the threshold of 67.2. 12 stage, or contribute less in the provision stage. 4 Experimental Procedure Experimental sessions were conducted in Vernon Smith Experimental Economics Laboratory (VSEEL) at Purdue University between April 2013 and September 2013. 140 subjects were recruited through ORSEE system (Greiner (2004)) and the experiment was programmed on z-Tree (Fischbacher (2007)) software. On average, the sessions lasted for about two hours and subjects’ earnings ranged from $9 to $45 (including a $4 participation fee) with the average of $20. All subjects were inexperienced in the sense that, no individual had participated in a previous CPR or public good experiments. The experimental instructions were given to the subjects in two steps. Upon arriving at the laboratory, subjects received the instructions for the earning activity. Subject were informed that their performance in the earning activity determined – with details on the exact amounts – the initial wealth they were going to receive at the beginning of each of the 20 experimental rounds. After the earning stage was completed and subjects’ wealth was realized to them, subjects were given the instructions for the rest of the experiment. Subsequently, subjects answered five incentivized questions that were designed to test their understandings of the instructions after which, the 20 rounds of provision-appropriation game followed. Each subject only participated in one experimental treatment. Upon completion of each session, subjects were privately paid in cash. 5 5.1 Experimental Results Contribution Decisions Prior to entering the appropriation stage, groups in the endogenous treatment decided how much to contribute from their wealth to the provision of the resource. Over the 20 rounds of the experiment, groups successfully met the provision threshold of 67.2 experimental dollars, about 68.85% of the times. The success rate increases to 71.54% over the last 10 rounds. 13 Recall that in the case of unsuccessful provision, contributions were refunded to the contributors. But, if group contribution exceeded the required threshold, the excess contribution was not rebated and was thus wasted. Conditional on successful provision, the waste decreases from 8.88 experimental dollars per period per group in the first five rounds, to 1.68 experimental dollars in the last five rounds. The decrease in the waste reveals that coordination among group members improves over time. Figure 1 graphs the average individual contribution (T̄i ) in the endogenous treatment by rich and poor players over time. Over the 20 rounds of the experiment, on average, rich players contribute about three times as much as poor players do in the provision process (T̄rich = 26.64 vs. T̄poor = 8.87). 0 5 Average Contribution 10 15 20 25 30 35 40 Average Contribution- Endogenous 0 5 10 Period Rich 15 20 Poor Figure 1: Average Contribution in the Endogenous Treatment Table 2 presents the results from a regression model that establish the significance of the difference between contributions by rich and by poor in the provision stage. The left column combines the data from all groups in the endogenous treatment. And the middle and the right columns report the effect for the groups that were successful and unsuccessful in meeting the required threshold separately. Since players’ roles are assigned by their performances in the earning activity, one might be concerned that the difference between contributions by rich and poor is originated from the difference in their ability to perform on the GMAT quiz.16 To control for the effect of ability, I include the quiz scores in the regressions as a regressor. 16 In other words, we may be concerned that they are the smarter (and not the rich) individuals who are making more cooperative contribution choices. 14 As the coefficients in the first row of the regression table confirm, rich players contribute significantly more towards the provision of the resource than poor players do. The higher contribution by rich cannot be explained by their ability to perform in the GMAT quiz. In fact the opposite holds true for the unsuccessful groups: while being rich increases the contribution in the provision process, better performance in the quiz, decreases the contribution 1 term indicates that for the unsuccessful groups. Finally, the positive coefficient for P eriod 17 contributions decrease over time. Table 2: Effect of Wealth on Individual Contributions in the Endogenous Treatment Regression Model (Random Effect) Dependent Variable: Individual Contribution All Groups Successful Groups Unsuccessful Groups Wealth Dummy 17.56*** 17.72*** 16.77*** (α) (2.31) (2.50) (1.97) 1 0.07*** 0.07*** 0.02 P eriod (0.02) (0.02) (0.29) (β) Quiz Score -0.54 0.05 -1.08* (γ) (0.63) (0.56) (0.60) Constant 9.23*** 8.01*** 10.47*** (δ) (1.81) (1.92) (2.00) Observations 988 684 304 Standard Errors are clustered at the group level and reported in parenthesis ∗p < 0.05, ∗ ∗ p < 0.01, ∗ ∗ ∗p < 0.001 1 Contributionit = δ + αW ealthDummyit + β P eriod + γQuizScorei + it The higher contribution by rich players is in line with the findings of previous threshold public good experiments with heterogeneous endowments (Rapoport and Suleiman 1993).18 As discussed in section 3, the higher contribution by rich in the endogenous treatment indicates that inequality aversion motivates subjects’ contribution choices. Result-1: In the endogenous treatment, rich players contribute significantly more towards the provision of the resource than poor players do. The difference in contributions between rich and poor is consistent with the predictions of inequality aversion theory. Using T1 (as opposed to T) allows us to account for higher variations in earlier periods, when learning is taking place with higher intensity. The result is robust to using either of the variations. 18 One important difference between the current experimental design and that of the threshold public good experiments is in the uncertainty in returns from contributions. In threshold public good experiments, payoffs only depend on whether or not the threshold is met (public good is provided), while here, the benefit from provision, not only depends on successful provision, it also depends on how individuals extract from the resource, once it is provided. In other words, there is more uncertainty about the benefits from contributions in this provision-appropriation experiment, which may lower the contributions. 17 15 5.2 5.2.1 Extraction Decisions Extractions by Type Inequality aversion and Equity theory predict different extraction pattern for rich and poor individuals under different provision mechanisms (Table 1). To explore whether these theories are able to explain the extraction choices in the experiment, I study individual extraction decisions by each type. Figure 2 graphs the average individual extractions by rich and by poor players, under each provision mechanism. In the regressive and progressive conditions, the average extraction by the poor players lies above that of the rich players in most rounds. To compare the extraction choices by different types, Table 3 reports the results from random effect Tobit regressions in which individual extractions are regressed on a wealth dummy, GMAT score 19 , group dummies and a term capturing the time trend. The first ,the second and the third columns report the regression coefficients for the regressive, the progressive and the endogenous treatments. 0 5 10 Period Rich 15 Poor 20 11 5 6 Average Extraction 7 8 9 10 11 Average Extraction 7 8 9 10 6 5 5 6 Average Extraction 7 9 8 10 11 12 Endogenous 12 Progressive 12 Regressive 0 5 10 Period Rich 15 20 Poor 0 5 10 Period Rich 15 Poor Figure 2: Average Extractions by Type The negative and significant coefficients of the wealth dummies in columns 2 and 3 indicate that in the regressive and the progressive treatments, rich players extract significantly less than poor players from the resource. This result holds, even after controlling for the performance in the GMAT quiz. Interestingly, not only the lower extraction by rich players is not explained by their ability, but the positive and significant coefficients of GMAT scores 19 Since the wealth assignments are not random and are determined by subjects’ performances in the GMAT quiz, I control for the performance in the quiz by including subjects’ GMAT scores as an independent variable in the regressions. 16 20 in columns 2 and 3 reveal that the opposite is true: controlling for the wealth levels, those subjects who performed better in the GMAT quiz extract more from the resource and not less. This observation is in line with the previous finding in the endogenous treatment that higher quiz scores predicted more self-centered contribution (lower Ti ) choices in unsuccessful groups.20 The higher extraction by poor in the regressive and the progressive treatments is consistent with the predictions of inequality aversion theory from section 3. Moreover, recall from section 3 that Equity theory predicted that in the progressive treatment, rich subjects should extract more than poor subjects and in the regressive treatment, we should not see any difference between extraction choices of rich and poor. However, as the regression results in Table 3 established, in both the regressive and the progressive treatments, rich players extract significantly less than poor players do. We thus conclude that Equity theory fails in correctly explaining subjects’ extraction choices in these two conditions. Result-2: In accordance with the predictions of inequality aversion, in the regressive and the progressive treatments rich subjects extract significantly less from the resource than poor subjects. No evidence exists that Equity theory motivates subjects’ extraction choices in these two conditions. As the regression coefficients in the third column of Table 3 reveal, there is no significant difference in extractions by rich and poor subjects in the endogenous treatment. Importantly, λ (the coefficient of individual contribution choice) is insignificant, which shows that even within each wealth group, those who contribute more in the provision stage, are not more likely to extract more in the appropriation stage. This finding leads us to reject Equity theory as a possible explanation for subjects’ choices in the endogenous treatment. Result-3: No significant difference exists in the extraction choices made by rich and poor players in the endogenous treatment. Equity theory cannot explain the subjects’ choices in the endogenous treatment. 20 The positive significant coefficient of the quiz score suggests that those who performed better in the quiz are more likely to make more selfish choices. This may be originated from the fact that individuals with higher scores are perhaps better at finding the equilibrium solutions, or it may merely be the case that the smarter individuals are in general more self-centered. The experimental design however, does not allow us to further disentangle such correlation. 17 Table 3: Effect of Wealth on Individual Extraction Decisions Random Effect Tobit Regression Dependent Variable: Individual Extraction (1) (2) (3) Regressive Progressive Endogenous Wealth Dummy -1.44** -1.62** 0.80 (α) (0.64) (0.75) (0.87) Quiz Score 0.28* 0.29* -0.40 (γ) (0.17) (0.20) (0.25) Condtribution -0.01 (λ) (0.02) 1 -0.02*** -0.05*** -0.03*** P eriod (0.01) (0.01) (0.01) (β) Constant 8.49*** 9.05*** 10.82*** (δ) (0.85) (1.06) (1.17) 1.26*** 1.52*** 1.57*** σu (0.17) (0.18) (0.17) 1.84*** 2.45*** 1.73*** σe (0.05) (0.07) (0.05) N 760 911 684 Standard Errors are reported in parenthesis , ∗p < 0.05, ∗ ∗ p < 0.01, ∗ ∗ ∗p < 0.001 1 Extractionit = δ + αW ealthDummyit + β P eriod + γQuizScorei + λContributionit + GroupDummies + it 5.2.2 Group Extractions and Earnings Across Treatments Figure 3 graphs average aggregate group extractions over time, for each treatment condi4 P tion.21 Average group extractions start slightly below the equilibrium prediction of xi = i=1 8 × 4 = 32 for self-interested agents and gradually increase over time. However, average group extractions in all treatments stays close to the equilibrium prediction for pure selfinterested agents in most periods. The non-parametric Wilcoxon signed-rank tests cannot reject the null hypothesis that the mean group extractions equal 32 in any of the treatments (p-value=0.72 for regressive; p-value=0.18 for progressive; p-value=0.52 for endogenous).22 Result-4: Consistent with the findings of the CPR literature, the aggregate group extractions are not significantly different from the Nash equilibrium prediction of the pure self-interested agents in any of the treatment conditions. 21 From the groups in the endogenous condition, only those who successfully provided the resource and actually made extraction decisions are included in the graph. 22 For the non-parametric analysis, I treat the average group extraction over all 60 rounds of the experiment for each group, as one independent data point. 18 20 22 Average group extraction 24 26 28 30 32 34 36 Average Group Extraction 0 5 10 Period Regressive Endogenous 15 20 Progressive Figure 3: Total Group Extraction in Each Treatment To compare the extraction choices across treatments, Table 4 reports the results from random effect Tobit regressions, in which aggregate group extraction is regressed on dummies for the regressive and the progressive treatments and a term capturing the time trend. The insignificance of the coefficients for treatment dummies indicate that overall, group extractions between the regressive and the endogenous, and between the progressive and the endogenous treatments are not significantly different from each other. The result of a Wald test reveals that aggregate extractions are not significantly different between the regressive 1 term and the progressive treatments either (p-value=0.26). The negative coefficient of P eriod confirms that group extractions increase over time. Table 4: Random Effect Tobit Regression (Dependent Variable: Aggregate Group Extraction) Random Effect Tobit 0.24 (1.02) Dummy for Progressive -0.92 (0.97) 1 -6.43*** P eriod (0.71) Constant 32.93*** (0.71) σu 2.22*** (0.32) σe 3.81*** (0.13) N 589 Standard Errors are reported in parentheses ∗p < 0.05, ∗ ∗ p < 0.01, ∗ ∗ ∗p < 0.001 1 + gt extractiongt = δ + αRegressiveg + βP rogressiveg + λ P eriod Dummy for Regressive 19 Result-5: No significant difference exists between aggregate group extractions across different provision mechanisms. Conditional on the provision of the resource, similar extractions across treatments would lead to similar earnings in the appropriation stage. However, taking into account the fact that – due to insufficient contributions – the provision process in the endogenous treatment is not always successful, the opportunity to benefit from the common resource occurs with a lower frequency in the endogenous treatment. This inefficiency leads to lower earnings from the appropriation stage in the endogenous treatment. Table 5 reports the results from random effect regressions, regressing group earnings on treatment dummies and a time trend. The first two columns drop the observations in the endogenous treatment in which the provision was unsuccessful, and the last two columns pool the data across all observations. In columns P 1 and 3, the dependent variable is the group earning from the appropriation stage ( Mi ), which includes the earning from private activity, plus the earning from extractions. And in columns 2 and 4, the dependent variable is the sum of group earning from the provision P P P P stage and the appropriation stage ( πi = Wi − Ti + Mi ). Table 5: Random Effect Regression- Group Earnings Across Treatments Dummy for Regressive Dummy for Progressive 1 P eriod Constant Only include the successful groups in the Endogenous treatment (1) (2) -1.35 2.51 (9.99) (10.03) 3.38 7.55 (9.52) (9.56) 51.64*** 47.00*** (6.98) (7.00) 185.55*** 475.35 6.89 (6.92) 21.63 21.72 37.61 37.74 0.25 0.25 589 589 Include both the successful and unsuccessful groups (3) (4) 57.81*** 39.71*** (15.46) (12.19) 62.53*** 44.75*** (14.71) (11.61) 46.13*** 42.13*** (10.60) (8.31) 127.42*** 439.07*** (10.38) (8.19) 34.02 26.87 60.52 47.43 24 0.24 665 665 σu σe ρ N P-value from t-Test H0 : EarningRegressive = 0.64 0.61 0.76 EarningP rogressive Standard Errors are reported in parenthesis * p < 0.05, ** p < 0.01, *** p < 0.001 1 groupearninggt = δ + αRegressiveg + βP rogressiveg + λ P eriod + gt 0.69 The insignificance of the coefficients of treatment dummies in columns 1 and 2 – along 20 with large p-values from the Wald tests reported in the last row – indicate that conditional on the successful provision of the resource, group earnings across treatments are not significantly different from each other. However not surprisingly, after including the cases of unsuccessful provision in the endogenous treatment (columns 3 and 4), group earnings in the regressive and the progressive treatments become significantly higher than the earnings in the endogenous treatment. 5.2.3 Extractions Choices as Responses to Extractions by Others To further explore subjects’ extraction decisions, this section studies individual decisions as a response to the choices made by group members. Specifically, I investigate whether the subjects respond to the choices of others, and if so, in what manner. To answer this question, Table 6 reports the regression results from random effect Tobit regressions in which individual extraction is regressed on previous period extractions by other individuals in the group by their types, along with a wealth dummy and a term capturing the time trend. The coefficients in the first and the second rows correspond to the previous period extractions by the other player of the same type, and the average previous period extractions by the two players of the opposite type, respectively. The first, the second, the third and the last three columns correspond to the responses by individuals in all treatments combined, in the regressive treatment, in the progressive treatment and in the endogenous treatment, respectively. In each three-column block, the first column pools the data for rich and poor players, and the second and the third columns reports the responses by the rich and by the poor players separately. The regression results reveal that in general, individuals do respond to the extraction choices made by others in their group, and they do so in an interesting manner. Note that all the significant coefficients in the first row are positive, while all the significant coefficient in the second row are negative. This implies that, individuals respond differently to the extraction choices made by the players of their own type, and the choices made by the players of the opposite type. Specifically subjects increase their extractions from the resource as the other player of their own type extracts more, while they decrease their extractions, as players of the opposite type extract more. The observed asymmetry in individual responses to their group-mates’ choices suggests that players hold different motives towards other individuals, depending on whether they happen to be of the same type, or of the opposite type. This 21 finding implies that individual utilities include others’ payoffs (or decisions) asymmetrically, and differentiate between payoffs (or decisions) of same-type and opposite-type players. Result-6: Subjects increase their extractions, as the other player of their type extracts more and decrease their extractions as players of the opposite type extract more from the resource. 22 23 All 0.02 (0.04) -0.05 (0.06) -0.70 (0.47) -0.05*** (0.01) 8.99*** (0.65) 1.42 (0.19) 1.82 (0.05) 720 -0.03*** (0.02) 6.94*** (0.80) 1.12 (0.20) 1.70 (0.07) 360 -0.06*** (0.02) 10.31*** (0.94) 1.76 (0.33) 1.92 (0.8) 360 Regressive Rich Poor 0.00 0.03 (0.05) (0.06) 0.12 -0.22*** (0.06) (0.09) All 0.04 (0.04) -0.14** (0.06) -0.72 (0.54) -0.08*** (0.02) 9.70*** (0.70) 1.76 (0.21) 2.44 (0.07) 864 -0.06*** (0.02) 7.35*** (0.83) 1.67 (0.28) 2.17 (0.08) 432 All 0.07** (0.04) -0.08** (0.04) -0.40 (0.52) -0.06*** (0.01) 8.71*** (0.41) 1.79 (0.20) 1.74 (0.05) 644 0.03 -0.11*** (0.03) 11.50*** (1.02) 1.82 (0.31) 2.70 (0.11) 432 0.01 Progressive Rich Poor 0.10* 0.00 (0.06) (0.06) -0.02 -0.30*** (0.07) (0.10) N P-values H0: 0.00 0.06 0.02 0.39 0.21 0.02 0.01 0.18 sametype Xt−1 othertype =X̄t−1 sametype Xt−1 :Extraction by the other player of the same type in the previous period othertype X̄t−1 : Average extraction by the two players of the opposite type in the previous period Standard Errors are reported in parenthesis. * p < 0.05, ** p < 0.01, *** p < 0.001 σe σu Constant 1 P eriod Wealth othertype X̄t−1 sametype Xt−1 All treatments All Rich Poor 0.05** 0.04** 0.04 (0.02) (0.03) (0.03) -0.07*** -0.04 -0.11*** (0.02) (0.03) (0.04) -0.60** (0.30) -0.06*** -0.05*** -0.07*** (0.01) (0.01) (0.01) 8.95*** 7.98*** 9.34*** (0.28) (0.31) (0.37) 1.68 1.58 1.79 (0.12) (0.15) (0.18) 2.04 1.82 2.26 (0.03) (0.04) (0.05) 2228 1114 1114 0.00 -0.04*** (0.02) 8.15*** (0.42) 1.81 (0.27) 1.42 (0.06) 322 0.61 -0.07*** (0.02) 8.82*** (0.48) 1.77 (0.29) 2.01 (0.09) 322 Endogenous Rich Poor 0.12*** 0.02 (0.04) (0.05) -0.13*** -0.03 (0.04) (0.06) Table 6: Random Effect Tobit Regressions- Responses to the Extraction Choices Made by Other Group Members 6 Discussion As discussed in section 3, the experiment has specific behavioral predictions from inequality aversion and Equity theories. The experimental results presented in section 5 established that the predictions of Equity theory are clearly refuted in all treatment conditions. In the regressive and the progressive treatments, in which subjects’ level of contribution in the provision process is set exogenously, higher contributions in the provision stage does not translate to higher extractions from the resource in the appropriation stage. In the endogenous treatment condition, in which users endogenously choose their level of contributions in the provision process, those who voluntarily contribute more in the provision stage do not extract more from the resource, once the resource is provided for their use. Thus, if users’ level of contribution in the provision process is exogenously determined by a central authority, through setting the level of taxes and fees that users pay, to cover the costs of providing and maintaining the common pool resource, the findings of this experiment suggest that using a provision rule that makes wealthier users pay a higher share of the costs does not necessarily lead them to extract more intensively from the resource as Equity theory suggests. Unlike the predictions of Equity theory, the results presented in section 5 showed that subjects’ choices in the experiment seem to be well aligned with the predictions of inequality aversion theory. Interestingly, subjects in the endogenous treatment act on their inequality aversion motives differently than those in the exogenous treatments: in the regressive and the progressive treatments, in which the level of contribution in the provision process is not subjects’ choice, inequality aversion motivates the extraction choices (rich subjects extract less than poor subjects do); and in the endogenous treatment, in which subjects get to voluntarily choose their level of contribution in the provision process, inequality aversion only motivates the contribution choices (rich subjects contribute 3 times as much as poor subjects do). To better understand why inequality aversion motivates the contribution choices in the endogenous treatment but the extraction choices in the two exogenous treatments, let’s compare subjects’ extraction choices in the progressive and in the endogenous treatments. Recall that in the progressive treatment, the entire (100%) provision cost was paid by the rich players, while in the progressive treatment, rich players pay only 75% of the provision cost (Result-1). Thus, when entering the appropriation stage, the degree of inequality between 24 rich and poor is higher in the endogenous than it is in the progressive treatment. If the degree of inequality was the only factor determining whether inequality aversion motivates the extraction choices or not, we should observe the difference between extractions by rich and by poor subjects to be higher in the endogenous treatment. Instead we observe no significant difference in extractions by rich and poor in the endogenous treatment but significantly higher extractions by rich players in the progressive treatment. Therefore, beside the degree of existing inequality, whether or not subjects had previously voluntarily taken action in reducing inequality seems to play an important role in determining if subjects are going to act on their inequality aversion motives in the appropriation stage. In other words, this observation suggests that the fact that subjects in the endogenous treatment have previously taken actions against inequality seems to reduce their discomfort about the current level of inequality.23 7 Conclusion This paper uses a laboratory experiment to explore how different ways of providing Common Pool Resources, affect their users’ extraction choices in a heterogeneous environment. Three experimental treatments are run in the lab through which, subjects either voluntarily (endogenous treatment) or mandatorily (regressive and progressive treatments) contribute to the provision of the resource, prior to making extraction decisions. To my knowledge, this is the first experimental study that compares CPR users’ choices under voluntary and mandatory provision schemes. It is also the first study that compares different provision mechanism in a heterogeneous CPR environment. Once the resource is provided, I don’t find any significant difference in aggregate group extractions across treatments. However, groups in the endogenous treatments fail to successfully provide the resource in about 30% of the times. The inability in successfully providing the resource in some groups leads to lower earnings, and causes the endogenous treatment to 23 This result is consistent with the findings of previous experiments about the role of intentionality in individual decision-makings. Blount (1995) finds that responders in the ultimatum games accept lower offers more often when offers are randomly generated than when they are chosen by another subject. Offerman (2002) finds that helpful (hurtful) actions are rewarded (punished) more frequently when they are intentional as opposed to when they are randomly generated. Charness (2004) finds that employees in a gift exchange game provide more(less) effort at high(low) wages when the wage is determined by the employer compared to when it is randomly generated. 25 be less efficient than the treatments in which contribution levels are determined exogenously. Inequality aversion and Equity theories, each predict different extraction patterns under each treatment condition. While I do not find any evidence for Equity theory motivating individual decisions, there exists clear evidence that inequality aversion guides choices: in the two treatments in which provision rule is determined exogenously, rich players reduce the initial inequality by extracting less than poor players from the resource. However, in the endogenous treatment, inequality aversion affects contribution – instead of extraction – decisions: rich players contribute three times as much as poor players do to the provision of the resource, and there is no difference in the extraction choices made by rich and poor. Studying individual responses to the choices made by others in their group reveals an asymmetry in players’ responses to their group members’ choices, depending on their type: individuals increase their extractions, as same-type players in their group extract more from the resource, and decrease their extractions, as opposite-type players extract more from the resource. While the results of any single experimental study should be regarded cautiously, the findings of this paper suggest several important policy implications. Many common pool resources are managed by government agencies who use tax revenues, collected from resource users to finance the provision and the maintenance costs. The presence of heterogeneity among resource users may create conflicting views about how the provision costs should be shared (e.g. Reuben and Riedl 2013; Nikiforakis, Noussair and Wilkening 2012). The findings of this paper show that, unlike the predictions of Equity theory, having the wealthier users pay a higher share of common resource provision costs does not make them act less cooperatively in extracting from it. Not only the progressive provision scheme did not lead to less optimal group outcomes, it decreased the inequality between rich and poor in this experimental environment. Of course, studying the effect of different of provision schemes on CPR users’ extraction choices in a laboratory environment is only a first step towards fully understanding how such policies affect individual decisions in the more complex field settings. 26 References Adams, J. S. (1963). TOWARD AN UNDERSTANDING OF INEQUITY. Journal of abnormal psychology, 67, 422436. Akpalu, W., & Martinsson, P. (2012). Ostracism and Common Pool Resource Management in a Developing Country: Young Fishers in the Laboratory. Journal of African Economies, 21(2), 266306. Arkes, H. R., Joyner, C. A., Pezzo, M. V., Nash, J. G., Siegel-Jacobs, K., & Stone, E. (1994). The Psychology of Windfall Gains. Organizational Behavior and Human Decision Processes, 59(3), 331347. Blount, S. (1995). When Social Outcomes Arent Fair: The Effect of Causal Attributions on Preferences. Organizational Behavior and Human Decision Processes, 63, 131144 Botelho, A., Dinar, A., Pinto, L. M. C., & Rapoport, A. (2013). Linking Appropriation of Common Resources and Provision of Public Goods Decreases Rate of Destruction of the Commons. NIMA Working Papers. Cardenas, J. C., Rodriguez, L. A., & Johnson, N. (2011). Collective action for watershed management: field experiments in Colombia and Kenya. Environment and Development Economics, 16(03), 275303. Cardenas, J. C., Stranlund, J., & Willis, C. (2002). Economic inequality and burdensharing in the provision of local environmental quality. Ecological Economics, 40(3), 379395. Cardenas, J.-C., & Ostrom, E. (2004). What do people bring into the game: experiments in the field about cooperation in the commons. CAPRi Working Papers. Cason, T. N., & Gangadharan, L. (2014). Promoting Cooperation in Nonlinear Social Dilemmas through Peer Punishment. Forthcoming in Experimental Economics Charness, G. (2004). Attribution and Reciprocity in an Experimental Labor Market. Journal of Labor Economics Cherry, T. L., Cotten, S. J., & Jones, L. R. (2013). The appropriation of endogenously provided common-pool resources. Resource and Energy Economics, 35(3), 329341. 27 Fehr, E., & Schmidt, K. M. (1999). A Theory Of Fairness, Competition, and Cooperation. Quarterly Journal of Economics, 114(3), 817868. Feld, L. P., & Frey, B. S. (2002). Trust breeds trust: How taxpayers are treated. Economics of Governance. Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10(2), 171178. Frey, B. S., Benz, M., & Stutzer, A. (2004). Introducing Procedural Utility: Not Only What, but Also How Matters. Journal of Institutional and Theoretical Economics JITE. Greiner, B. (2004). The Online Recruitment System ORSEE 2.0 - A Guide for the Organization of Experiments in Economics. Working Paper Series in Economics. Hackett, S. C. (1992). Heterogeneity and the Provision of Governance for Common-Pool Resources. Journal of Theoretical Politics, 4(3), 325342. Hackett, S., Schlager, E., & Walker, J. (1994). The Role of Communication in Resolving Commons Dilemmas: Experimental Evidence with Heterogeneous Appropriators. Journal of Environmental Economics and Management, 27(2), 99126. Hayo, B., & Vollan, B. (2012). Group interaction, heterogeneity, rules, and co-operative behaviour: Evidence from a common-pool resource experiment in South Africa and Namibia. Journal of Economic Behavior & Organization, 81(1), 928. Hoffman, E., McCabe, K., Shachat, K., & Smith, V. (1994). Preferences, Property Rights, and Anonymity in Bargaining Games. Games and Economic Behavior, 7(3), 346380. Homans, G. C. (1961). Social behavior: Its elementary forms. (R. K. Merton, Ed.)NY Harcourt Brace Javanovich (p. 404). Harcourt, Brace & World. Janssen, M. A., Anderies, J. M., & Joshi, S. R. (2011). Coordination and cooperation in asymmetric commons dilemmas. Experimental Economics, 14(4), 547566. Konow, J. (2003). Which Is the Fairest One of All? A Positive Analysis of Justice Theories. Journal of Economic Literature, XLI(December), 11881239. 28 Margreiter, M., Sutter, M., & Dittrich, D. (2005). Individual and Collective Choice and Voting in Common Pool Resource Problem with Heterogeneous Actors. Environmental Resource Economics, 32(2), 241271. Marks, M., & Croson, R. (1998). Alternative rebate rules in the provision of a threshold public good: An experimental investigation. Journal of Public Economics. Nikiforakis, N., Noussair, C. N., & Wilkening, T. (2012). Normative conflict and feuds: The limits of self-enforcement. Journal of Public Economics, 96, 797807. Offerman, T. (2002). Hurting hurts more than helping helps. European Economic Review, 46, 14231437 Ostrom, E., Gardner, R., & Walker, J. (1994). Rules, games, and common-pool resources. (E. Ostrom, Ed.)The University of Michigan Press Ann Arbor MI TECHNOLOGY SOCIETY ENVIRONMENT (Vol. 6, p. 369). University of Michigan Press. Rapoport, A., & Suleiman, R. (1993). Incremental Contribution in Step-Level Public Goods Games with Asymmetric Players. Organizational Behavior and Human Decision Processes, 55(2), 171194. Reuben, E., & Riedl, A. (2013). Enforcement of contribution norms in public good games with heterogeneous populations. Games and Economic Behavior, 77, 122137. Solstad, J. T., & Brekke, K. A. (2011). Does the Existence of a Public Good Enhance Cooperation among Users of Common-Pool Resources? Land Economics, 87(2), 335345. Sutter, M., Haigner, S., & Kocher, M. G. (2010). Choosing the Carrot or the Stick? Endogenous Institutional Choice in Social Dilemma Situations. Review of Economic Studies, 77(4), 15401566. Thaler, R. H. (2008). Mental Accounting and Consumer Choice. Marketing Science, 27(1), 1525. Trawick, P. B. (2001). Successfully Governing the Commons: Principles of Social Organization in an Andean Irrigation System. Human Ecology, 29(1), 125. Villena, M. G., & Zecchetto, F. (2011). Subject-specific performance information can worsen the tragedy of the commons: Experimental evidence. Journal of Economic Psychology, 32(3), 330347. 29 Wahl, I., Muehlbacher, S., & Kirchler, E. (2010). The impact of voting on tax payments. Kyklos: International Review for Social Sciences, 63(1), 144158. 30
© Copyright 2026 Paperzz