Paretian dictators: constraining choice in a voluntary contribution game

Robert J.Oxoby

Constitutional Political Economy ISSN 1043-4062 Volume 24 Number 2 Const Polit Econ (2013) 24:125-138 DOI 10.1007/s10602-013-9139-6

1 23

Your article is protected by copyright and all rights are held exclusively by Springer Science +Business Media New York. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com.

1 23

Author's personal copy

Const Polit Econ (2013) 24:125138 DOI 10.1007/s10602-013-9139-6 ORIGINAL PAPER

Paretian dictators: constraining choice in a voluntary contribution game

Robert J. Oxoby

Published online: 17 May 2013 Springer Science+Business Media New York 2013

Abstract We explore individuals preferences over limiting the choice sets of others in an environment with externalities. Specically, we conduct public goods games in which participants can mandate the contributions of others or restrict choices to a subset of feasible contributions levels. We nd that, relative to a baseline treatment in which individuals make choices from the set of all contribution alternatives, allowing individuals to constrain the choices of others results in more efcient outcomes. We discuss these results in light of the literature on behavioral theories of reciprocity and conditional cooperation and in regards to the literature on pre-constitutional design, political institutions, and social choice. Keywords Public goods Conditional cooperation Reciprocity Experiments C7 C9 D4 H4

JEL Classication

1 Introduction The tension between allowing unfettered individual choice and implementing efcient outcomes lies at the heart of the PigouvianCoasian debate inherent in the literature on public goods provision and constitutional design. The fundamental question here rests in the nature of restrictions on choice and how these restrictions map into benets enjoyed by a populace. Indeed while some have argued that individuals must face restrictions of liberties and choices in order to preserve social
R. J. Oxoby (&) Department of Economics, University of Calgary, 2500 University Drive NW, Calgary, AB T2N1N4, Canada e-mail: oxoby@ucalgary.ca R. J. Oxoby Canadian Institute for Advanced Research, Toronto, ON, Canada

123

Author's personal copy

126 R. J. Oxoby

order, others (notably Hayek 1955; 1960) have argued for unrestricted choices and the universal and equal enforcement of basic rights (e.g. property rights) in order to yield efcient outcomes. This suggests that there are fundamental issues regarding how individuals value their own and others rights (e.g. property rights, freedom of choice) that are entangled in the implementation of economic and social policies. One approach to understanding how individuals value others choices and rights has been through theories of fairness (see Konow 2003) in which individuals have preferences over their own and others allocations (e.g. fairness or inequity aversion), the processes yielding these allocations (e.g. cooperation and reciprocity), and the entitlements of others to certain allocations (e.g. procedural justice and equity theory). These theories suggest that individuals value others rights and that ideals such as fair treatment and conditional cooperation may mitigate the conict between self-interest and efciency characterizing many economic environments (e.g. the provision of public goods). However, while these theories have found support in laboratory and eld experimentation, there are often difculties reconciling theories of social preferences with observed behaviors. For example, policies which one might consider to be supported based on theories of social preferences or equality of opportunity (e.g. welfare transfers, redistribution, afrmative action) are hotly debated, indicating individuals differing views on how these policies affect outcomes.1 These debates largely focus on who reaps the costs and benets of a proposed policy and how individuals will behave in the presence of a policy, demonstrating that there exists signicant uncertainty regarding how different decision-makers will behave before the same policy. The key in efcient policy design may therefore lie in creating institutions in which individuals must act similarly in the face of a policy, potentially by constraining the choices available under certain policy instruments. This suggests that individuals may prefer different policies depending on how others choices or behaviors are limited, implying (as in the constitutional design literature) that individuals may be willing to trade-off freedom of choice for the implementation of efcient outcomes. In this paper, we explore the issues of freedom of choice and equal application of rules by considering how individuals choose to constrain or mandate others decisions in a simple experimental environment. We conjecture that, as predicted by theories of social preferences (particularly those embodying reciprocity and conditional cooperation), individuals will make choices that are more efcient when they can eliminate or reduce uncertainty regarding the behavior of others. Moreover, as argued in the literature on political and legal institutions, individuals will prefer rules that are universally and equally applied to all individuals over rules favoring an individual or group (even if they are a member of the favored group). The use of a laboratory experiment to address these issues permits a stark analysis regarding the tensions between individuals decisions, uncertainty, and efciency in the provision of public goods.
1

For example, while redistribution may be supported by theories of social preferences (e.g. Fehr and Fischbacher 2002), a large literature (e.g. Tullock 1959; Usher 1981) suggests that redistribution may lead to instability.

123

Author's personal copy

Paretian dictators 127

To this end, we utilize a linear public goods game in which individuals can dictate or restrict the choices of others in their group. That is, individuals can either dictate the contributions of other participants or can limit their contribution options to a subset of potential contributions. Our results yield two insights: First, individual display a preferences for outcomes that maximize total material payoffs (i.e. Pareto efcient outcomes based on preferences over wealth): individuals dictate efcient (i.e. maximal) contributions on the part of other participants and prefer restricting others (and their own) choices to high level contributions.2 Secondly, individuals display a preference for conditional cooperation or contribution matching: Even when individuals can dictate different contribution levels for themselves and others, a majority of participants choose symmetric, optimal contributions for both themselves and others. Other research, particularly Cettolin and Riedl (2011) and Fischbacher and Gaechter (2008), have demonstrated that conditional cooperation is not enough to motivate high contributions in public goods games. Rather, individuals have a preference to contribute less than others and, in some cases, respond to coercion. In our setting, restricting the choice of others can be construed as a form of coercion in terms of mandating the contribution choices of others. This ability reduces (and in some cases eliminates) the opportunity to contribute less than other participants, thereby providing a means to implement high contribution levels. This suggests that if an individual can dictate that others behave in a fully cooperative manner (here, contribute total wealth to the public good), there is a preference to behave in a similar manner even though maximizing private material payoffs predicts free-riding (i.e. zero contribution). The experimental literature on public goods games has focused to a large extent on voluntary contribution games (particularly linear public goods games; see Ledyard 1995 and Zelmer 2003). Within the linear public goods literature, two stylized facts that emerge: (i) individuals rarely play their Nash strategies (i.e., minimal contributions) and (ii) individuals rarely achieve Pareto efcient outcomes (i.e., all participants making maximal contributions).3 This suggests that while individuals have preferences over their own and others wealth as suggested by models of altruism and inequity aversion (e.g. Andreoni 1990; Fehr and Schmidt 1999; Goeree et al. 2002), their behavior is tempered by concerns over others behavior. Indeed, evidence suggests that contributions in public goods games are strongly inuenced by concerns over reciprocity, characterized by either contribution matching/conditional cooperation (Croson et al. 2005; Fischbacher et al. 2001; Keser and van Winden 2000) or the threat of negative reciprocity (i.e. punishment chter 2000; 2001).4 by others for free riding; Fehr and Ga

Throughout the paper we refer to the outcome yielding joint wealth maximization as the Pareto efcient outcome. Such would be the case in environments with transfers. For discussions of behaviors in other types of public goods games, see Holt and Laury (2005). Other factors affecting contributions in linear public goods games include the marginal product of contributions, group size, and differences in wealth levels (Buckley and Croson 2006; Chan et al. 1996; Cherry et al. 2005; Isaac et al. 1994). Other research has suggested that contributions are driven by decision errors (e.g. Anderson et al., 1998; e.g. Goeree et al. 2002). Additionally, there is evidence that differences in the source of individuals endowments (and hence differences in individuals perceived

3 4

123

Author's personal copy

128 R. J. Oxoby

Under our treatments in which participants can restrict the choice sets of others (and themselves) we nd efcient outcomes are frequently implemented: We observe over 80 % of participants choosing the Pareto efcient outcome (i.e. that which maximizes total payoffs) when they can dictate the choices of others, suggesting that efciency can be achieved in these environments when the uncertainty surrounding others behaviors can be reduced. This has implications for the structure of collective choice mechanisms, suggesting that individuals may coordinate on an efcient outcome if they are able to limit the extent to which others may shirk or free ride. Thus, when working at the level of a social planner, Pareto efciency emerges as a criterion for equilibrium selection (cf. Charness and Rabin 2002). This has important implications for our understanding of institutional design and our results shed light on issues of constitutional design and endogenous political institutions (e.g. Buchanan and Tullock 1962; Aghion et al. 2004; Kroll et al. 2007; Voigt 1997) in which there exist incentives to ex ante constrain others choices in order to implement optimal outcomes. The remainder of the paper is organized as follows: Sect. 2 describes the experiment; results are presented in Sect. 3. Section 4 discusses our results in light of theories of choice and constitutional design. Section 5 concludes.

2 Experimental design Our experiment follows a one-shot, linear public good game. In each treatment, participants were endowed with ten dollars and randomly and anonymously assigned into four-participant groups. In our baseline treatment participants made simultaneous contribution decisions ci between zero and ten dollars (integer values).5 In each treatment, an individuals payoffs were given by Pi  Pci ; ci 10 ci 0:4
4 X j 1

cj :

Our treatments were designed with an eye towards examining how individuals made choices when they could explicitly determine or signicantly constrain the contributions made by others in a public goods environment while still maintaining a degree of freedom over their own choice. For this we implemented three treatments where an individuals choice varied between to the extent in which they could determine or constrain the choices of others. In our symmetric dictator (SD) treatment participants were asked to choose a contribution value cAll between zero and ten dollars which would be applied to all participants in their group. In the asymmetric dictator (AD) treatment, each

Footnote 4 continued property rights) affect allocation decisions (e.g. see Cherry et al. 2002; Hoffman and Spitzer 1985; Hoffman et al. 1994; Konow 2000; Oxoby and Spraggon 2008).
5

Instructions available upon request.

123

Author's personal copy

Paretian dictators 129

participant chose a contribution level for herself cOwn and a contribution level for all other individuals in her group cOthers. In each of these treatments, once all participants had made their decisions a dictator was randomly chosen and her contribution choices were implemented within her group. In our quasi dictator (QD) treatment each participant was asked to provide a menu of three possible contribution levels {cA, cB, cC} from which participants in her group would choose (where cA \ cB \ cC). That is, each participant was asked for a constrained set of three contribution levels from which all members of her group could decide (thereby constraining group members decisions from a set of eleven elements to a set of three elements). After all participants made their choices regarding the three-element menu, one individual was randomly selected and her menu was presented to members of her group. Individuals then made simultaneous contribution choices based upon this menu, ci 2 fcA ; cB ; cC g: In the baseline treatment, individuals made simultaneous decisions and, after all participants made their decisions, payoffs were revealed. In each of the dictator treatments, we employed a strategy vector method in which individuals made choices prior to knowing whether or not they would be in the role of the dictator. That is, all participants in a given treatment made decisions over cAll (SD treatment), {cOwn, cOthers} (AD treatment) or {cA, cB, cC} (QD treatment) prior to knowing whether theirs would be the role of dictator. After all participants made their choices, one individuals choice was randomly selected and implemented. In the case of the SD and AD treatments, this determined participants payoffs; in the QD treatment, the menu {cA, cB, cC} was implemented and participants made simultaneous contribution choices ci 2 fcA ; cB ; cC g: These choices then determined participants payoffs. Note that given four-participant groups, each individual faced a 25 % probability that her initial choice would be implemented.6 After payoffs were revealed, participants completed a short demographic survey. A brief comment is warranted on our use of a single-shot public goods game. Although many public goods experiments use repeated interactions to permit participants to learn the game or coordinate within their groups, a large literature has focused on individual behavior in one-shot public goods games (see, among chter and Riedl 2005, Goeree et al. 1999, Kocher others, Fischbacher et al. 2001, Ga et al. 2008, and Walker and Halloran 2004). Our choice to use a one-shot public good game was based on trying to provide the starkest environment in which constrained choice could affect decision making. To this end, we used a one-shot game to avoid aspects of reputation building, signalling, or direct reciprocal actions that ow from repeated public goods games (e.g. the analysis in Spraggon and Oxobby 2009).

While some experiments nd no difference in behavior resulting from the use of this elicitation procedure (e.g. Brandts and Charness 2000; Oxoby and McLeish 2004) other have found that the use of the strategy method affects behavior (e.g. Brosig et al. 2003). Note that our method is not strictly the strategy method (individuals are not making a schedule of decisions); rather each individual makes a decision and one individuals decision is randomly chosen and implemented.

123

Author's personal copy

130 R. J. Oxoby

2.1 Predictions As a benchmark, consider the Nash prediction based on preferences only over an individuals own wealth. In the baseline treatment, the Nash prediction is for participants to fully free ride, opting for a zero contribution ci = 0 and each obtaining Pi 10: In the SD treatment, since the dictator can remove most uncertainty regarding the contribution of others, the wealth maximizing choice corresponds to the efcient choice of cAll = 10 yielding Pi 16 for all. On the other hand, wealth maximization in the AD treatment implies choosing maximal contributions for all others and zero contribution for oneself: {cOwn, cOthers} = {0, 10} yielding Pi 22 for the dictator and Pi 12 for all other individuals in the group. Finally, in the quasi-dictator treatment, a participants wealth maximizing menu is to restrict all participants to contributing at the upper interval of the choice set, {cA, cB, cC} = {8, 9, 10}. Given this menu, each agent should individually choose the lowest element, cA = 8 and obtain Pi 14:80: In each treatment, the efcient outcome (i.e. that maximizing aggregate wealth) is realized when each individual contributes ci = 10.

3 Results One hundred and twenty-eight individuals participated in the experiment, yielding eight four-person groups per treatment (32 participants per treatment). Each session consisted of a single treatment and included 16 participants (i.e., two sessions for each treatment yielding eight sessions). Participants were drawn from the undergraduate student body at the University of Calgary and recruited using the software developed by Greiner (2004). The experiments were conducted in a dedicated experimental economics laboratory at our university and programmed in z-Tree (Fischbacher 2007). Table 1 presents the summary statistics of elicited contributions for all participants by treatment. Participants earned an average of $14.59 (r = 1.93) across treatments. Of our participants, 63 % were male and 78 % were members of the Faculty of Arts, with the largest single major being economics (36 %). We conducted analyses based on the demographic information but identied no differences in behavior based on gender or area of study. Nonparametric Wilcoxon test results for comparisons of individual contributions across treatments are presented in Table 2. Note that contributions in the three
Table 1 Average contributions and payoffs (standard deviations) by treatment

Treatment Baseline Symmetric dictator Asymmetric dictator

n 32 32 32 32

Average contribution 3.47 (2.11) 9.25 (1.67) 9.31 (1.87) 8.59 (2.14)

Average payoff 12.08 (1.99) 15.55 (1.00) 15.58 (1.15) 15.16 (1.48)

All amounts denominated in dollars

Quasi-dictator

123

Author's personal copy

Paretian dictators Table 2 Nonparametric Wilcoxon p values for individuals contributions in each treatment 131

Baseline Baseline SD AD

SD \0.01

AD \0.01 0.2862

QD \0.01 0.1298 0.0229

dictator treatments are all signicantly different from those in the baseline treatment.7 The difference we identify between the AD and QD treatments (while nding no difference between the SD and AD treatment nor the SD and QD treatments) appears to be driven by the fact that only one group in the AD treatment had a dictator that did not choose efcient contributions for themselves and all others. As a result, 7 of 8 groups in the AD treatment implemented the Pareto efcient outcome. Note that the character of play in each of the dictator treatments is fundamentally different. That is, in each dictator treatment, individuals made different types of choices regarding their behavior and that of others. As such we cannot directly compare the initial choices made by participants in the SD, AD, and QD treatments. However, relative to the baseline treatment, each of the dictator treatments yield similar results. Table 3 presents the average (standard deviation), median and modal choices in the SD, AD, and QD treatments. That 27 of 32 participants in the SD treatment (84 %) chose the maximum contribution of ten dollars demonstrates that individuals recognize the Pareto efcient outcome and know how to implement this outcome. Note that this behavior conforms with the wealth maximizing Nash prediction in this treatment. Perhaps more importantly, 22 of 32 (68.75 %) of participants chose Pareto efcient contributions in the AD treatment, demonstrating that not only do individuals recognize the efcient outcome, but also that once participants have means to mitigate others free-riding (here, by dictating others contributions), their own contributions are consistent with implementing the Pareto efcient allocations. We nd that 68.75 % of participants were willing to give up $6 in favor of behaving in a cooperative manner and matching others maximal contributions. This result is consistent with theories of contribution matching and conditional cooperation (Croson et al. 2005; Fischbacher et al. 2001; Keser and van Winden 2000), preferences for inequity aversion (e.g. Fehr and Schmidt 1999) and efcient outcomes (as discussed in Charness and Rabin 2002) and Kantian decision making as discussed by Laffont (1975) and Bilodeau and Gravel (2004) in which the universalization of moral actions result in efcient outcomes. That only one individual chose the Nash contributions (cOwn, cOthers) = (0, 10) while 78 % of
7

Such high observed contributions are similar to pubic goods games with communication such as Cason and Khan (1999) and Isaac and Walker (1988) where communication between subjects results in higher contribution levels which (in the latter) are maintained and approach the efcient provision of the public good. In a related experimental design, Cason and Mui (1997) nd that, consistent with Social Comparison Theory, teams of two dictators make offers which display a greater regard for others payoffs than single dictators. This suggests that communication, even in a sub-sample of the population, results in greater cooperative behavior.

123

Author's personal copy

132 Table 3 Summary statistics for choices in the SD, AD, and quasi-dictator treatments R. J. Oxoby

Treatment SD AD cAll cOwn cOthers QD cA cB cC

Mean 9.28 (2.11) 7.94 (3.44) 9.25 (2.09) 7.50 (1.50) 8.75 (0.80) 10.00 (0.00)

Median 10 10 10 8 9 10

Mode 10 (84.38) 10 (68.75) 10 (84.38) 8 (87.50) 9 (87.50) 10 (100.00)

Numbers in parentheses are standard deviations. Numbers in brackets are the percentage of respondents choosing the modal value

participants chose cOwn C 5 demonstrates how the potential breakdown in contribution matching inuences the relatively low contributions observed in standard voluntary contribution games (e.g. our baseline treatment). It is interesting to note that 27 of 32 participants in the AD treatment chose contributions levels such that cOwn 2 fcOthers 2; cOthers 1; cOthers g: This suggests that ideas of contribution matching and reciprocity drive individuals to make contributions which are close to those of others. In our quasi-dictator treatment, 28 of 32 participants (87.5 %) followed the Nash strategy and chose to constrain individuals choices to {cA, cB, cC} = {8, 9, 10}, demonstrating that participants recognize the Pareto efcient outcome and constrain themselves and others to choosing contributions close to the efcient contribution. Indeed, every individual in the quasi-dictator treatment chose cC = 10, indicating their recognition of the potential for efcient contributions. Moreover, when faced with this constrained choice set, 68 % of participants chose ci = 10 while only 18 % chose ci = 8.8 Table 4 presents the average earnings in each treatment. In line with the contributions reported in Table 1, earnings are higher in the dictator treatments relative to the baseline treatment. Moreover, notice that the distribution of payoffs in dictator treatments are relatively less unequal than in the baseline treatment: Table 4 presents the average Gini coefcient for each treatment (average of the Gini coefcients for each group in a treatment) and average inequality in each treatment relative to the baseline treatment. Given contributions matching, there is
It is interesting to note that two groups in the QD treatment had implemented menus of {cA, cB, cC} = {5, 8, 10} and {cA, cB, cC} = {4, 7, 10}. In these two groups, average contributions were 5.00 (r = 2.00) and 5.75 (r = 1.5). This suggests that when the implemented menu left open the possibility for substantial free-riding (i.e. a value of cA signicantly less than cC), participants interpreted this as a signal that the quasi-dictator (i.e. the individual who chose the implemented menu) would freeride. This result is akin to the menu dependency discussed by Sen (1997) and identied in the experiments th et al. (2001) in which less fair outcomes are rejected less when more of Bolton et al. (2005) and Gu equal allocations are unavailable. From a signalling perspective, this result is similar to the behaviors of decision makers in Van Huyck et al. (1993) in which auction bids served as signals of individuals intended strategies.
8

123

Author's personal copy

Paretian dictators Table 4 Payoff statistics in each treatment Treatment Average payoff 12.08 15.55 15.58 15.16 r Avg. Gini coefcient 0.088 0.000 0.004 0.046 Inequality Rel. to baseline (%) 100 0.0 4.5 52 133

Baseline Symmetric dictator Asymmetric dictator Quasi-dictator

1.99 1.00 1.15 1.48

The average Gini coefcient in a treatment is the average of the Gini coefcients for each group within the treatment

substantially less inequality in the dictator treatments.9 Thus, if agents are inequity averse (as in Bolton and Ockenfels 2000; as in Fehr and Schmidt, 1999) this points to another aspect in which outcomes in the dictator treatments Pareto dominate those in the baseline treatment.

4 Discussion Perhaps the most curious aspect about behavior in linear public goods environments is not that individuals contribute non-zero amounts, but that they do not contribute more.10 Indeed, theories of social preferences based on inequity aversion (e.g. Fehr and Schmidt 1999), reciprocity (e.g. Dufwenberg and Kirchsteiger 2004; Falk and Fischbacher 2006), and encompassing efciency concerns (e.g. Charness and Rabin 2002) suggest that individuals have reasons to contribute more than the dominant strategy zero-contribution. Our results suggest that obtaining an efcient allocation is a desire of individuals, but that concerns over the contributions of others reduces ones own contributions to the low levels frequently observed in linear public goods games. As such, one way to interpret the linear public goods game is as an assurance game in which relatively low contributions are the result of participants (who may have preferences over reciprocity or conditional cooperation) hedging their contributions in the face of uncertainty regarding the behavior of others. In our experiments, when participants have a means to mandate contribution matching we observe individuals choosing contributions which implement the Pareto efcient outcome. That we observe 68 % of participants choosing ci = 10 when restricted to choosing a contribution level in {8, 9, 10} suggests that restricting others choice sets may be sufcient to create assurances which motivate efcient contributions. As such, our results build on those of Fischbacher and Gaechter (2008) who identify individuals preference to contribute less than other. In our setting, this ability is largely eliminating, permitting a means for groups to implement high contributions from all participants.
9

Alternate measures of inequality (e.g. Atkinsons index) yield similar results regarding the relative inequality across the treatments. Andreoni (1988) suggests that individuals may choose larger-than-Nash contributions strategically in repeated public goods environments.

10

123

Author's personal copy

134 R. J. Oxoby

Moreover, as shown by Cettolin and Riedl (2011), conditional contribution is often not enough to garner high contributions. Our implemented means of restricting others choices buttresses motives of conditional contributions, thereby enabling Pareto efcient outcomes even in our quasi-dictators treatment. Alternately, one may interpret this result as arising from reciprocity (as suggested in Dufwenberg and Kirchsteiger 2004) or, perhaps more appropriately, anticipatory reciprocity (Cherry et al., 2005): In choosing a set {8, 9, 10} the dictator has not only restricted the contributions of others, but also her own contribution. This may serve as a signal of her behavior in which she anticipates others to behave in a reciprocal manner. Note that the observed outcomes not only satises Pareto efciency based on aggregate wealth maximization, but also yields an outcome consistent with conditional cooperation and inequity aversion.11 In terms of policy, our results are in spirit to that arising under rebate rules for threshold public goods. Such rules are often utilized to for the provision of public goods or in the cases of motivating charitable contributions (see the discussions in Marks and Croson 1998). Experimentally, Isaac et al. (1989) and Marks and Croson (1998) nd that such rules increase voluntary contributions. Our results also suggest that individuals may be amenable to rules mandating certain behaviors from themselves and others. While our experiments considered only a dictator choosing contribution levels (or potential levels), note that a voting mechanism would have yielded similar results in our environment. That is, given the results in Table 3, voting over group contribution levels would have yielded outcomes very similar to those observed. Thus our results lend experimental support for the provision of referenda which assist in mandating the behavior of others in environments where individuals have a desire to coordinate on efcient outcomes. Indeed our results suggest that a publicly selected provision mechanism may yield greater cooperation by restricting individuals to match one anothers contributions. These restrictions are similar to those arising in the context of communication and face-to-face interactions in which individuals actions are constrained (perhaps socially) in such a way to reduce transaction costs and the need for costly monitoring (Cason and Khan 1999). This in turn facilitates the functioning of development programs and the provision of local public goods. Our results have implications for issues surrounding pre-constitutional design. For example, as following (see Voigt, 1997), suggesting that individuals have preferences in which they are willing to restrict others choices as a means to implementing more efcient outcomes, suggesting that Pareto efciency may serve as an equilibrium selection device when individuals can make choices at a constitutional level.12 That
11

From the stance of equity theory (Walster et al. 1978), these outcomes are also efcient in that the inputs individuals have contributed are proportional to the outputs they receive. The behavior in our environment may also be reective of the norms a decision maker thinks she and others should abide by, as suggested by the Kantian norms discussed in Bilodeau and Gravel (2004). In terms of limiting choices, Boudreaux and Lipford (1998) use a Buchanan-Tullock framework to argue for less inclusive voting rules in the European Union. The central idea here is that as group size and heterogeneity increases, less inclusive voting rules reduce costs. In our environment, less inclusive rules over contribution levels reduce the cost of uncertainty over others behavior, resulting in in more efcient outcomes. It should be noted that the work of Kocher et al. (2011) has demonstrated that risk preferences alone are not sufcient to explain contributions in public good games.

12

123

Author's personal copy

Paretian dictators 135

is, in our environment where the strategic interdependence of individual choices may result in Pareto inferior outcomes, individuals are able to restrict the choices of themselves and others and thereby constrain the set of possible outcomes. From a constitutional perspective, this is akin to mandating certain behaviors on the part of individuals (here, contributions) in order for individuals to enjoy improved outcomes (here, Pareto efcient payoffs). As such, our research falls in line with that of Kroll et al. (2007) in which the focus is on the efciency implications of voting rules with and without punishment. As purely constitutional approach (e.g. one in which participants can choose whether to have a symmetric, asymmetric, or quasi-dictaor) remains an area of future research. Moreover, as discussed by Hayek (1955, 1960), we observe individuals demonstrating a preference for equality under the rule of law (particularly in our AD treatment) in which all individuals choices (including those of the dictator) are constrained in the same manner. That is, when given the opportunity to discriminate between themselves and others, participants eschewed an asymmetric, individually wealth maximizing set of group contributions in favor of an equally applied contribution rule. This preference is manifest through almost 70 % of participants choosing symmetric contributions such that cOwn = cOthers = 10. Thus, a majority of participants chose to abide by the contribution rules they would have imposed on others.13 Relatedly, Aghion et al. (2004), Harstad (2005), and others emphasize the importance of super-majorities and majority rules in legislation and constitutional design. To this end, it is interesting to note that in each of our dictator treatments, the efcient outcome was chosen by a super majority, suggesting that (when incentives are appropriately aligned) efcient outcomes receive the lions share of support. In our AD treatment, this also implies that a (super) majority of individuals preferred an outcome in which the rule of law was equally applied to all decision makers. Just as Harstad (2005) argues that majority rules can be designed to internalize externalities and implement efcient outcomes, our results demonstrate that universally applied restrictions on individuals choice sets can also be used to implement efcient outcomes. This suggests an important trade-off between liberty (i.e. unlimited choice) and democracy in which individuals may democratically support restrictions in choice when such restrictions facilitate implementing efcient allocations. Our results may also shed light on the tension between Pareto efciency and liberty as discussed by Sen (1970, 1976). In our simple setting, individuals constrained the choices of others (and themselves) when these constraints facilitated implementing efcient allocations.14 Finally, the results of our experiments suggest that limiting individuals choices (particularly when individuals are concerned about coordination or matching) may yield superior outcomes. This falls in line with the research of Iyengar and Lepper (2000) who nd that individuals subjective
13 A possible interpretation of this result is in broad terms as evidence that individuals prefer equality in treatment, as in before the law or by the government. 14 Our experiment does not consider individuals willingness to give up some choices nor the value assigned to freedom of choice discussed by Hayek (1960) and Sen (1976).

123

Author's personal copy

136 R. J. Oxoby

evaluation of a chosen good is lower when the choice was made from a larger set than from a smaller set.15 By restricting the choices of others (even to a single event), uncertainty and strategic interdependence (i.e. the potential for free-riding) are signicantly reduced in our dictator treatments. Although not true in all decision environments, in our setting such restrictions on choice sets creates the potential for more efcient and more preferred outcomes to be realized.

5 Conclusion How individuals anticipate and account for the behavior of others when making decisions is a fundamental question in game theory and decision analysis. Theories of reciprocity and conditional cooperation postulate that individuals behavior is highly sensitive to what they think others will do. In this paper we report on a public goods experiment in which individuals had the opportunity to either make decisions for others or restrict others contribution choices to a subset of previously available choice alternatives. We nd that when individuals can mandate or restrict the choices of others, they do so in a manner which facilitates the realization of Pareto efcient outcomes. From the standpoint of pre-constitutional design, our results suggest that individuals may be willing to trade liberty (in the form of freedom of choice via larger choice sets) for increased facility in implementing socially superior outcomes.
Acknowledgments We thank Robin Boadway, John Boyce, Francisco Gonzalez, Erin Krupka, Michael McKee, Kendra McLeish and John Spraggon for suggestions. Financial support was provided by the Social Sciences and Humanities Research Council of Canada and the Canadian Institute for Advanced Research.

References
Aghion, P., Alesina, A., & Trebbi, F. (2004). Endogenous political institutions. Quarterly Journal of Economics, 119(2), 565611. Anderson, S. P., Goeree, J. K., & Holt, C. A. (1998). A theoretic analysis of altruism and decision error in public goods games. Journal of Public Economics, 70, 297323. Andreoni, J. (1988). Why free ride? Strategies and learning in public goods environments. Journal of Public Economics, 70, 291304. Andreoni, J. (1990). Impure altruism and donations to public goods: A theory of warm glow giving. Economic Journal, 70, 464477. Bilodeau, M., & Gravel, N. (2004). Voluntary provision of a public good and individual morality. Journal of Public Economics, 88(34), 645666. Bolton, G. E., Brandts, J., & Ockenfels, A. (2005). Fair procedures: Evidence from games involving lotteries. The Economic Journal, 115, 10541076. Bolton, G. E., & Ockenfels, A. (2000). A theory of equity, reciprocity, and competition. American Economic Review, 30(1), 166193. Boudreaux, K., & Lipford, J. (1998). Group size, group heterogeneity, and voting rule: An application of the Buchanan-Tullock model to the European Union. European Journal of Law and Economics, 5, 133152. Brandts, J., & Charness, G. (2000). Hot vs. cold: Sequential responses and preference stability in experimental games. Experimental Ecnoomics, 2, 227238.
15 In a similar spirit, Sethi-Iyengar et al. (2004) nd that participation in 401(k) plans is greater when there are fewer option choices.

123

Author's personal copy

Paretian dictators 137

Brosig, J., Weimann, J., & Yang, C.-L. (2003). The hot versus cold effect in a simple bargaining experiment. Experimental Economics, 6(1), 7590. Buchanan, J., & Tullock, G. (1962). The calculus of consent. Ann Arbor, MI: University of Michigan Press. Buckley, E., & Croson, R. (2006). Income and wealth heterogeneity in the voluntary provision of linear public goods. Journal of Public Economics, 90(45), 935955. Cason, T. N., & Khan, F. U. (1999). A laboratory study of voluntary public goods provision with imperfect monitoring and communication. Journal of Development Economics, 58(2), 533552. Cason, T. N., & Mui, V.-L. (1997). A laboratory study of group polarisation in the team dictator game. Economic Journal, 107(444), 146583. Cettolin, E., & Riedl, A. (2011). Partial coercion, conditional cooperation, and self-commitment in voluntary contributions to public goods. Technical report. Chan, K. S., Godby, R., Mestelman, S., & Muller, R. A. (1996). Spite, guilt and the voluntary provision of public goods when income is not distributed equally. Canadian Journal of Economics, 29(2), S605 S609. Charness, G., & Rabin, M. (2002). Understanding social preferences with simple tests. Quarterly Journal of Economics, 117(3), 817869. Cherry, T. L., Frykblom, P., & Shogren, J. F. (2002). Hardnose the dictator. American Economic Review, 92(4), 12181221. Cherry, T. L., Shogren, J. F., & Kroll, S. (2005). The impact of endowment heterogeneity and origin on public good contributions: Evidence from the lab. Journal of Economic Behavior and Organization, 57(3), 357365. Croson, R., Fatas, E., & Neugebauer, T. (2005). Reciprocity, matching and conditional cooperation in two public goods games. Economics Letters, 87(1), 95101. Dufwenberg, M., & Kirchsteiger, G. (2004). A theory of sequential reciprocity. Games and Economic Behavior, 47(2), 268298. Falk, A., & Fischbacher, U. (2006). A theory of reciprocity. Games and Economic Behavior, 54(2), 293315. Fehr, E., & Fischbacher, U. (2002). Why social preferences matterThe impact of non-selsh motives on competition, cooperation, and incentives. The Economic Journal, 112, C1C33. chter, S. (2000). Cooperation and punishment in public goods experiments. American Fehr, E., & Ga Economic Review, 90(4), 980994. chter, S. (2001). Fairness and retaliation: The economics of reciprocity. Journal of Fehr, E., & Ga Economic Perspectives, 14(3), 159181. Fehr, E., & Schmidt, K. (1999). A theory of fairness, competition, and cooperation. Quarterly Journal of Economics, 114(3), 817868. Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10(2), 171178. chter, S., & Fehr, E. (2001). Are people conditionally cooperative? Evidence from a Fischbacher, U., Ga public goods experiment. Economics Letters, 71, 397404. Fischbacher, U., & Gaechter, S. (2008). Social preferences, beliefs, and the dynamics of free riding in public good experiments. Technical report. chter, S., & Riedl, A. (2005). Moral property rights in bargaining with infeasible claims. Management Ga Science, 51(2), 249263. Goeree, J. K., Holt, C. A., & Laury, S. K. (1999, September). Altruism and noisy behavior in one-shot public goods experiments. Virginia Economics online papers 331, University of Virginia, Department of Economics. Goeree, J. K., Holt, C. A., & Laury, S. K. (2002). Private costs and public benets: unraveling the effects of altruism and noisy behavior. Journal of Public Economics, 83(2), 255276. Greiner, B. (2004). The online recruitment system ORSEE 2.0a guide for the organzation of experiments in economics. Working paper series in Economics 10, University of Cologne. th, W., Huck, S., & Mueller, W. (2001). The relevance of equal splits in ultimatum games. Games and Gu Economic Behavior, 37, 161169. Harstad, B. (2005). Majority rules and incentives. Quarterly Journal of Economics, 2005(120), 15351568. Hayek, F. A. (1955). The political ideal of the rule of law. Cairo: National Bank of Egypt. Hayek, F. A. (1960). The constitution of liberty. Chicago: University of Chicago Press.

123

Author's personal copy

138 R. J. Oxoby

Hoffman, E., McCabe, K., & Smith, V. L. (1994). Preferences, property rights, and anonymity in bargaining games. Games and Economic Behavior, 7, 346380. Hoffman, E., & Spitzer, M. (1985). Entitlements, rights, and fairness: An experimental examination of subjects concepts of distributive justice. Journal of Legal Studies, 14, 259297. Holt, C. A., & Laury, S. K. (2005). Theoretical explanations of treatment effects in voluntary contributions experiments. In C. Plott & V. Smith (Eds.), Handbook of experimental economics results, Vol. 1. New York: Elsevier Press. Isaac, R. M., & Walker, J. M. (1988). Communication and free-riding behavior: The voluntary contribution mechanism. Economic Inquiry, 26(4), 585608. Isaac, R. M., Walker, J. M., & Schmidtz, D. (1989). The assurance problem in a laboratory market. Public Choice, 62, 217236. Isaac, R. M., Walker, J. M., Williams, A. W. (1994). Group size and the voluntary provision of public goods: Experimental evidence utilizing large groups. Journal of Public Economics, 54(1), 136. Iyengar, S. S., & Lepper, M. R. (2000). When choice is demotivating: Can one desire too much of a good thing?. Journal of Personalilty and Social Psychology, 79(6), 9951006. Keser, C., & van Winden, F. (2000). Conditional cooperation and voluntary contributions to public goods. Scandanavian Journal of Economics, 102(1), 2339. Kocher, M. G., Martinsson, P., Matzat, D., & Wollbrant, C., (2011, January). The role of beliefs, trust, and risk in contributions to a public good. Working papers in Economics 482, University of Gothenburg, Department of Economics. Kocher, M. G., Martinsson, P., & Visser, M. (2008). Does stake size matter for cooperation and punishment?. Economics Letters, 99(3), 508511. Konow, J. (2000). Fair shares: Accountability and cognitive dissonance in allocation decisions. American Economic Review, 90(4), 10721091. Konow, J. (2003). Which is the fairest one of all? A positive analysis of justice theories. Journal of Economic Literature, 41(4), 11861237. Kroll, S., Cherry, T. L., & Shogren, J. F. (2007). Voting, punishment, and public goods. Economic Inquiry, 45(3), 557570. Laffont, J.-J. (1975). Macroeconomic constraints, economic efciency and ethics: An introduction to kantian economics. Economica, 42(168), 43037. Ledyard, J. O. (1995). Public goods: A survey of experimental research. In J. H. Kagel & A. E. Roth (Eds.), Handbook of experimental economics. New Jersey: Princeton University Press. Marks, M., & Croson, R. (1998). Alternative rebate rules in the provision of a threshold public good: An experimental investigation. Journal of Public Economics, 67(2), 195220. Oxoby, R. J., & McLeish, K. N. (2004). Sequential decision and strategy vector methods in ultimatum bargaining: Evidence on the strength of other-regarding behavior. Economics Letters, 84(3), 399405. Oxoby, R. J., & Spraggon, J. (2008). Yours and mine: Property rights in dictator games. Journal of Economic Behavior and Organization, 65, 703713. Sen, A. K. (1970). The impossibility of a Paretian liberal. Journal of Political Economy, 78, 152157. Sen, A. K. (1976). Liberty, unanimity, and rights. Economica, 43, 217245. Sen, A. K. (1997). Maximization and the act of choice. Econometrica, 65, 745779. Sethi-Iyengar, S., Huberman, G., & Jiang, W. (2004). How much choice is too much? Contributions to 401(k) retirement plans. In O. S. Mitchell & S. Utkus (Eds.), Pension design and structure: new lessons from behavioral nance. Oxford: Oxford University Press. Spraggon, J. M., & Oxobby, R. J. (2009). Game theory for playing games: Sophistication in a negativeexternality experiment. Economic Inquiry, 47(3), 467481. Tullock, G. (1959). Problems of majority voting. Journal of Political Economy, 67, 571579. Usher, D. (1981). The economic prerequisite to democracy. New York: Columbia University Press. Van Huyck, J. B., Battalio, R. C., & Beil, R. O. (1993). Asset markets as an equilibrium selection mechanism: Coordination failure, game form auctions, and tacit communication. Games and Economic Behavior, 5(3), 485504. Voigt, S. (1997). Positive constitutional economics: A survey. Public Choice, 90, 1153. Walker, J. M., & Halloran, M. A. (2004). Rewards and sanctions and the provision of public goods in oneshot settings. Experimental Economics, 7(3), 235247. Walster, E., Walster, G., & Berscheid, E. (1978). Equity: Theory and research. Boston: Allyn and Bacon. Zelmer, J. (2003). Linear public goods experiments: A meta-analysis. Experimental Economics, 5, 299310.

123