Common Pool Resource with Free Mobility: Experimental Evidence from the US and Mongolia Dolgorsuren Dorj October 27, 2009 Abstract The Tragedy of Commons may become worse if people are free to move between different commons. We conducted laboratory experiments in US and Mongolia with the common pool resource setting in which people freely chose between two localities that may differ by governing structures: no regulation or sanctions. Governing structures were imposed either exogenously or chosen by majority voting. We found that under free mobility conditions efficient resource use is still attainable under the sanctions if the target harvesting level adjusts to the population level. People self-select into different regimes based on their behavioral types. JEL classification codes: C7, C72, C91, Q2, R12. Keywords: common pool resource, free mobility, sanctions, experiment

1

Introduction

To date, the majority of field and theoretical studies on commons have focused on separate communities. In these studies, decisions made by a particular member regarding resource extraction only affect his/her community. In reality, however, in and out population migration may affect resource use across communities. For example, world-wide intense use of the resources brought environmental challenges such as habitat loss, bycatch, pollution affecting more than single community. Little is known about the effect of migratory pressure and patterns of behavior in such mobile commons. Most of the previous research on common pool resources (Hardin, 1

1968, Ostrom, 1990) and on the use of various sanctioning systems in commons (Casari and Plott, 2003; Visser, 2006; Velez, Murthy and Stranlund, 2009) concern with the over-exploitation problem within one community. However, Ostrom (1990) and others highlight importance of clear defined boundaries to successful self-governance in the commons. We relax assumption of no entry and consider situtations when the commons are prone to population pressure. Our work differs from previous research in that it addresses the mobility issue by looking at multiple localities each with a common pool resource (CPR hereafter). We study how free mobility between communities affects resource use. Population migration is free in a Tiebout sense: we assume that mobility is costless. We compare resource extraction behavior under different resource use regimes (sanctions vs. no sanctions) using laboratory experiments across two countries: USA and Mongolia. Examples of commons with multiple locations include pastureland under extensive grazing or international fisheries. Extensive livestock production prevalent in many countries such as Mongolia, Kyrgyzstan, Niger, Mali relies on low-forage-value grassland which requires movement of herders along larger territory subsuming many localities. Similarly, fisheries around international waters involve resource use problem across many localities or even countries. For example, pink salmon in the NE Pacific is harvested in Canadian, American and Russian waters. Interestingly, some localities may have regulations such as sanctions which promotes efficient use of the resource while others may be unregulated as regulations can be introduced gradually. For example, Alaskan salmon fishery in Chignik was governed in certain times by cooperative fleet that shares catch equally among members and in other times by independent fleet in such a way that unregulated fishermen were free to join cooperatives (Deacon et al. 2003). To reflect the high mobility aspect of such setting we extend the existing research to allow for two communities in which citizens of each area are free to choose the place to extract the resource. In each community, there is a territory with resources for harvesting activity. We consider two possible community management regimes for each locality: (i) unregulated community with no rules towards the grazing activity; (ii) sanctioning mechanism with mutual monitoring as studied in Casari and Plot (2003) that resembles community property regime. The research question is how does free mobility affect the performance of the sanctioning system? If one locality is regulated and the other is not, would sanctioning institution withstand the migratory pressure from the unregulated locality? How does the difference in management regime affect agent’s decision to join one of the localities? We address these questions in an experimental two-community environment CPR framework, where participants make decisions regarding the location to graze and the harvesting level. 2

The objective of this paper is to study experimentally efficiency of resource use in CPR game with free mobility assuming either no sanctions or specific sanctioning mechanism within localities. Case studies from Ostrom (1990) suggest that graduated sanctions used against the excessive use of the resource are a helpful tool in CPR environments. However, in a laboratory setting, Walker et. al (1990) report that costly sanctions do not alter the result of classical Prisoner’s Dilemma game, and the resource is used above the selfish Nash equilibrium if monitoring is a costly activity for the selfish players. In CPR experiments with communication alone, with sanctions alone, or with communication and sanctioning opportunity, Ostrom et al. (1992) find that repeated communication improves per person net benefit. They show that with no institutions, resource is destroyed quickly. Gardner et al. (1997) demonstrate that neither entry restrictions nor quota caps increase efficiency above the Nash prediction. In contrast, revenue generating monitoring in Casari and Plott (2003) as part of a special system “Carte de Regola” which was used to manage the common properties in Alpine villages may bring the outcome close to the social optimum. Schmitt et al. (2000) examine CPR when only a subset of decision-making group had opportunity of face-to-face communication with each other and find that uncertainty about the outsider’s decision decreases the efficiency of the resource use. Walker et al. (2000) find that proposals and simple majority voting over allocation rules raise efficiency of commons. Unlike the above, in our experiments subjects do not propose the regime to be used; participants simply vote for the available institutions. Experiments by Vyrastekova and van Soest (2003) emphasize that participant’s voting in favor of appropriate incentive scheme improves the efficiency of the resource use. In this experimental study we focus on the classical CPR model and consider two possible regimes: no regulaion and the sanctioning mechanism of Casari and Plott (2003). Our contribution is to add free mobility to the model. Our aim is to (i) study whether individual’s behavior changes in response to changes in the resource governing structure; (ii) test whether the sanctioning system can stand the migratory pressure from unregulated locality and may improve the welfare across two localities. In addition we consider effects of exogenous vs. endogenous institutions on resource use under free mobility conditions. This comparison allows us to consider differences between representative democracy and constitutional democracy. In some real life settings, in the short-run horizon, institution may be imposed top-down by community’s representatives, or by the head of organization, or by minority that rules the corporation on behalf of owners. In this sense the representative democracy provides institutions that are exogenously given. On other hand the direct participation in the institution selection by mean of voting resembles en3

dogenous formation of institution1 . We study whether the choice of institutions out of exogenously given alternatives may be supported by the majority voting rule. More specifically, we test whether “voting with the feet”in a Tiebout (1956) sense and “voting with the ballot” conditions produce similar results. By “voting by feet” we mean the condition where citizens choose between communities, given their exogenous institutions while “voting with the ballot” means that citizens vote for the regime in each community by majority rule. A natural question to ask is: given a choice between sanctions and no sanctions which type of institution will people prefer? Would people self-select according to their behavioral types in two different localities in the exogenous setting? Would subjects vote for sanctions when both localities are free to have either sanctions or no sanctions in the endogenous setting? It may happen that some communities will choose efficient institution while others do not. Banfield (1958) explains the cross-country growth differences by community’s ability to successfully handle the commons problem, by ability to build networks beyond the family, and by presence of political associations and corporate organizations. He points that the lack of successful associations, membership and active citizenship to be one of the most limiting factor to economic development. The relationship between voting behavior and resulting efficiency of organization has also been subject to experimental studies in the public good context. Gurerk et al. (2006) show how over time entire population achieves cooperation by shifting from sanctions-free society to the sanctions institution in the VCM where voters choose between sanctions and no sanctions option. Sutter et al. (2006) study exogenous vs. endogenous institutions in VCM and find that rewards option is chosen more often than the punishment option and most subjects in partner matching prefer institutional arrangement without punishment. Botelho et al. (2005) find no punishment as a voting outcome when subjects choose between punishment or no punishment at the final round after experiencing both institutions in the VCM, strangers setting. Earlier studies in public good setting (Sefton et al., 2002; Fehr and Gachter, 2002) and in the bargaining game (Andreoni et al., 2003) find reward structure less effective than punishment mechanism. Our paper differs from above studies in several aspects. First, we consider CPR setting. Second, while above studies are mostly interested in the preferred institution people would choose from available alternatives, we focus on the spatial distribution of population in two-communities economy, and the effect of one community’s regime on the other community’s performance. As 1

In our endogeneous institutions setting participants only choose the preferred regime from available alternatives by majority rule. However, they do not choose the strength of the punishment regime as in Ertan et al. (2005) or select voting rules endogenously as in Decker et al. (2003)

4

Sutter et al. (2006) we consider ”voting by ballot” in addition to ” voting by feet” present in Gurerk et al. (2006). However, we used the majority rule in the voting stage rather unanimity rule. In our setting voting is a costless decision and subjects can’t abstain from voting. If our paper consider voting procedure in each round of the session; in Sutter et al. (2006) participant’s have only one-time voting procedure at the beginning of session effective for all ten consequent rounds. Our experiments show that: (a) under free mobility conditions, efficient resource use is still attainable in communities which adopt sanctions provided that the target harvesting level adjusts to the population level; (b) subjects “vote with the feet” for sanctioning system over unregulated regime so that more people locate in the regulated (sanctioning) locality; (c) partial monitoring is sufficient to induce higher level of efficiency under the sanctioning regime. These results are explained by the majority of subjects who rationally respond to institutional incentives and cooperate in a regulated locality, but overuse the resource in an unregulated community. We call such people “institution-responsive” types. At the same time we find that there are significant minorities who are less sensitive to institutions. We call “free riders” those subjects who tend to over-harvest irrespectively of institution, and “cooperators” those who restrain harvesting at or below socially optimal level regardless of institution. Regarding these behavioral types, we find that there is a sorting of population: free riders cluster in the unregulated locality and overuse the resource; in contrast, cooperators and institution-responsive types gather in the regulated locality keeping efficiency at high level. Further, when given a chance to choose a regime by majority rule subjects “vote with the ballot” for and against sanctioning system depending on their behavioral types; cooperators and institution-responsive types vote for sanctions most of the time whereas free riders vote for no regulation at the beginning. However, through competition free riders learn that sanctions are a better governing structure. Despite differences in the economic, political and cultural background across two countries, the results are consistent across US and Mongolia subjects except some differences in voting behavior. This result is partly due to the student population exhibiting a homogenous behavior across countries. The next two sections provide predictions of CPR model with and without free mobility. Section 4 explains experiment design. Section 5 reports the results of the experiment with and without free mobility. In section 6 we discuss and conclude.

2

Basic CPR model

We briefly review theoretical predictions for cases of interest. 5

2.1

One locality, No Sanctions

We use a standard CPR model as in Falk et al. (2002). The baseline commonpool resource game assumes no regulation. A finite number of agents N , each with endowment e, simultaneously decide on the amount of appropriation xi , where P i ≤ N denotes the agent’s index. Let X = N i=1 xi be the total appropriation and total revenue f (X) be a concave production function2 . The cost per unit of harvest denoted by c is common and independent of every other agent’s decision while the revenue for each agent will depend on harvesting choice of all agents. Then each agent i0 s profit is given by πi = e − cxi + [xi /X]f (X). The symmetric Nash equilibrium appropriation by each agent it follows (Falk et al., 2002): Proposition 1 If the locality is unregulated, denoted by U , then the Nash equilibrium total appropriation X U = Nn+1 · a−c is higher than the social optimum, b Xsopt = (a − c)/2b, if N > 1. The equilibrium per person profit is πiU = e +

(a−c)2 . b·(N +1)2

For example, suppose total population in the locality is N = 5. The parameters of production function are as follows: a = 14.5, b = 1/30, c = 2.5. We use this parameters in our experimental setting. Then, equilibrium total appropriation is 300 compared to social optimal of 180, with equilibrium per person appropriation of 60 and a profit of 120 (Table 1).

2.2

One locality, Sanctions

The model with sanctions (Casari and Plott, 2003) assumes that community restricts per agent harvesting to threshold amount, λ which targets socially optimal harvesting level. First, each agent chooses a use level. Second, after observing total appropriation all agents are free to monitor each other. By paying monitoring fee, k, any agent may obtain exact information about the harvesting decision of one other member. In case of multiple inspections only one person is randomly chosen as inspector and receives the direct transfer from monitoring. Harvesting beyond the threshold costs individual a fine payment if any other member of community discovers his/her violation. For each excess harvesting unit a violator pays unitary fine, h. The fine the violator pays is a transfer to the inspector who discovers the violation. We use superscript R for the regulation or sanctions. The two stage game is solved by backward induction and the following is established in Casari and Plott (2003): 2

For simplicity, we assume f (X) = aX − bX 2 if X ≤ 435, f (X) = 200 · (e(−0.0575(X−435)) − 1) if X > 435.

6

Proposition 2 Suppose a locality has a sanctioning mechanism, where the threshold is set as λ = xopt −k/h−ε, where ε > 0 is small enough and xopt = X opt /N . Further, let per unit fine be given by h = a − c − xopt (N + 1) · b. Then this mechanism supports the socially optimal level of harvesting as the subgame perfect Nash equilibrium, X R = X opt . Also in this equilibrium everyone inspects each other, with perceived inspection probability being equal to one, p∗ = 1. 2

(a−c) The equilibrium per person profit is given by πiR = e + b·(N − k. Note that +1)2 R U for any given population size, N > 1, πi (N ) < πi (N ) as long as the monitoring cost k is small enough3 . Therefore, each agent prefers the sanctioning regime to no regulation. Efficiency level in community is equal to 100 percent if the monitoring cost equals zero. However, it is less than 100 percent if we account for the monitoring Q Q Q Q Q cost. Efficiency is defined as: E = ∗ / opt ·100 = ( −n · e)/ opt ·100, where ∗ Qopt is the sum of actual profits minus sum of endowment while indicates maximum social surplus. To continue with an example let the parameters of production function be the same as before N = 5, a = 14.5, b = 1/30, c = 2.5 and parameters of sanctioning system be λ = 34, k = 7, h = 4.8. Then total appropriation with sanctions X R = X opt = 180 is smaller than without sanctions. Per person harvest declines to 36 and per person profit of 209 is much higher than under no regulation.

3

CPR with Free Mobility

To study how free mobility affects resource use we now extend the models to two communities. Consider two communities with identical production function, f (X). Let the total population in two localities be N . We assume free mobility, which means that migration is costless.

3.1

Exogenous institutions and free mobility: Sanctions in one locality and No sanctions in the other locality

Consider two neighboring communities. First, we assume that regulatory regimes in each locality are exogenous. Let one locality have sanctions and the other locality have no regulation at all. We assume that the harvesting threshold in a regulated locality is set optimally and adjusts instantaneously to the population level in the 3

The monitoring cost has to satisfy k ≤

(a−c)2 b

7

1 · ( 4·N −

1 (1+N )2 ).

locality: λ = λ(N R ), defined as in Proposition 2. Further, the inspection cost is low enough so that the benefits of sanctioning mechanism outweigh the costs of adopting it in a regulated locality for any population level n ≤ N . Consider free mobility equilibrium in a game, where agents are free to move from one location to another. First, each agent chooses the locality (S j ), j = 1, 2 to live. Then once harvesting threshold in the regulated community is adjusted according to the population level, eaach agent i decides on a harvesting level (xi ) within the chosen locality. Further, each agent in a regulated locality may inspect other agents after observing total group use. All discovered violations become public information. Then fines and payoffs are realized. In order to obtain predictions with free mobility across two localities we use the notion of Tiebout equilibrium (Greenberg and Weber, 1986) often referred as a free mobility equilibrium, partition of population of agents into localities, where no single agent wants to move from the current position to join other existing locality. In the free mobility equilibrium, two conditions must hold for each locality: (i) all localities are inhabited and agents’s actions are optimal within each locality; (ii) no agent wants to move, i.e. each agent’s profit in a chosen locality is at least as high as the profit in the other locality. For our purposes, we need to add dynamic structure to the model. Hence, we solve for the free mobility equilibrium as the subgame perfect Nash equilibrium of the game where first stage involves location choice, the second stage involves harvesting decisions followed by monitoring decisions if sanctions the case. We assume that agents first choose the localities in an anticipation of the no regulation CPR game in locality with no regulation and anticipation of the sanctioning mechanism in the regulated locality. Thus, agents split into two localities so that per agent profits across two localities are equalized. This leads to larger population in the locality with sanctions. Proposition 3 In the free mobility equilibrium with asymmetric institutions and identical agents, the locality with the sanctioning system accommodates more individuals than the unregulated locality, N R > N U . The appropriation levels satisfy X opt = X R < X U . The sanctioning system introduced in one locality improves the welfare of all agents in both localities as compared to no regulation in either locality. To provide an example suppose the parameters of the CPR with free mobility game are set as N = 10, a = 14.5, b = 1/30, c = 2.5, λ = 28, k = 7, h = 5 4 . The unregulated locality accommodates nU = 4 agents and regulated locality inhabited by 4

In the locality with sanctions, the threshold λ and fines, h, varied with the population level as Table 2 in Part 3 of Instructions. From Proposition 2 a threshold (λ) was adjusted to the population size as follows: no threshold if n=1 ; λ=87 if n=2; λ=58 if n=3; λ=43 if n=4; λ=34 if

8

nR = 6 agents. However, total appropriation in the unregulated locality X U = 288 is greater than appropriation in the regulated locality X R = X opt = 180. Interestingly, efficiency in the localities both with and without sanctions improves from 55.6 percent with no regulation to 96.1 and 64 percent respectively. Note that sanctions in one locality create a positive externality for the unregulated neighboring locality by reducing it’s population pressure through agents migrating to the regulated locality, and welfare increases in both localities as compared to the no regulation case.

3.2

Endogenous institutions and free mobility: Voting Equilibrium

In this part we relax the assumption of exogenous institutions. We allow each community to choose their own governing institution using majority voting (Fiorina and Plott, 1978). We show that the regulatory regime can be sustained in both localities in a subgame perfect Nash equilibrium under majority voting. Again, each individual chooses first a locality. Next, each member of the community votes either for sanctions or no sanctions in their locality, and the outcome is determined by majority voting. Third, each agent decides on an individual harvesting level. In the communities with sanctioning regime, monitoring decisions follow and the payoffs are then realized. Again we use subgame perfect Nash equilibrium concept to solve for equilibrium with voting. Proposition 4 In the voting equilibrium, agents vote for sanctions in both localities. The resource is used at the efficient level. The appropriation levels across localities are the same, XR1 = XR2 = X opt and the population sizes are identical, NR1 = NR2 . This result is obtained using the median voter theorem (Duncan, 1948; Downs, 1957). Since in our model all agents are identical and share homogeneous preferences, by the median voter theorem the outcome of the majority voting in each locality is the median voter’s preferred institution, that is, sanctions as profits under sanctions are higher. To follow with an example suppose the parameters of the CPR free mobility game with voting are N = 10, a = 14.5, b = 1/30, c = 2.5, λ = 34, k = 7, h = 4.8. n=5; λ=28 if n=6; λ=24 if n=7; λ=21 if n=8; λ=18 if n=9; λ=16 if n=10. From Proposition 2 a fine (h) varied with the population size in the following way h=0 if n=1 ; h=3 if n=2; h=4 if n=3; h=4.5 if n=4; h=4.8 if n=5; h=5 if n=6; h=5.14 if n=7; h=5.25 if n=8; h=5.33 if n=9; h=5.4 if n=10. See Casari (2005) on fine-to-fee ratio.

9

Then equilibrium split of the population is n1 = n2 = 5. Both localities choose sanctions as a regime and appropriation in each of them are X1 = X2 = X opt = 180. Both localities achieve high efficiency, 96.8 percent with per person profit of 209.

4

Experimental Procedures and Design

Computerized experiments involved 40 undergraduates and graduate students from the University of Hawaii at Manoa (USA) and 80 students from the Academy of Management (Mongolia) in May-June 2007. Ten participants were invited to each session. Subjects were seated separately from each other at laboratory computer terminals. The instructions were provided in verbal and written format. See the instructions in the Appendix of paper. No communication was allowed. The payoffs were in terms of “francs” 5 . Each subject was paid in cash at the end of the session. Subjects received 5 dollars for participation and earned on average additional 17.5 dollars in the USA (US hereafter) site, while the average payoff in Mongolia (MG hereafter) site was 12.5 dollars. Sessions lasted less than two hours. We conducted four treatments in two designs (See Tables 1 and 2). In each session, several treatments were implemented sequentially. Design 1 included three treatments: no mobility and no sanctions (N -treatment ), followed by no mobility and sanctions (S-treatment), followed by free mobility and asymmetric institutions (N S-treatment), where one market had no regulation and other market had sanctions. The first two treatments investigated the effect of the institution (sanctions) on the efficiency of resource use. The last treatment (N S-treatment) tested the effect of free mobility on the institution’s performances. Note that the institutions in the NS-treatment are exogenous, where subjects “vote with the feet”, choosing their preferred locality in a Tiebout sense. The second design had four treatments: no mobility and no sanctions (N), followed by no mobility and sanctions (S), followed by free mobility and asymmetric institutions (NS), followed by free mobility and voting (V). In addition to the three treatments described above, in V-treatment subjects voted for an institution in the chosen market. There were two markets labeled “A” and “B”, and subjects initially chose one of the two markets. Having observed the number of participants in his/her market, each subject voted for an institution, and the institution was chosen by majority vote. After the voting stage, the chosen institution was announced in each locality and investment decisions followed. The voting treatment was added to capture the differences between “voting 5

Laboratory artificial currency that was converted into domestic currency, either US dollars or Mongolian tugrugs, at the end of each session.

10

with the feet” and “voting with the ballot” conditions. Within-subjects design was chosen to let the subjects experience both N and S conditions before the two-market free mobility NS-condition. The sequence of treatments replicated historical realities, where tragedy of commons have been experienced first (N-treatment), and the regulatory mechanism often followed as a solution to the free rider problem (Streatment). Also, we pursued a structure that allowed subjects to start with simple instructions before going on to more complex environments. A subject’s decision consisted of the following actions: (i) choice of a market in the free mobility treatments only; (ii) voting decision in the voting treatment only; (iii) investment decision; (iv) inspection decision in markets with sanctions. Instructions were read aloud to everyone at the beginning of each treatment. After the instructions were read, a quiz was given before each treatment to ensure that they understood the instructions clearly. Subjects had two practice periods to become familiar with the software. At the end of the session a questionnaire was administrated which provided feedback on the strategy the subjects followed. Each period, the computer screen displayed a history table of the previous plays within subject’s own group only (for one market treatments) or within both markets (for free mobility treatments). We now describe specific procedures for each treatment.

No sanctions in one market (N -treatment) In N -treatment, ten subjects were randomly placed in two groups of five subjects and were told they will remain in the same group throughout the duration of the treatment. The partner-matching reflects long term interaction among herding families in one community. Each subject had 10 tokens per period as endowment. In each period, each participant had an opportunity to invest between 1 to 500 tokens in their market. Each token that the subject ordered cost him c = 2.5 experimental francs. Gross group return and return on tokens invested were presented in Table 1 of instructions (part 1, N). Also each subject was provided with a more detailed table, where total group investment increased by single unit increments. Subjects were informed about the patterns of return in the market which depended on the total investment. After each of total eight rounds subjects receive a feedback on total investment, return per token in the market, and his/her payoff.

11

Sanctions in one market (S-treatment) The same group of five subjects as in N-treatment then experienced the sanctions treatment. Once an investment decisions were made, each participant could inspect others. Monitoring per person required 7 francs, while each token discovered above the threshold transferred 4.8 francs from violator to the inspector. If many people asked to monitor a particular person, one of them was randomly chosen as an inspector; all others were treated if they did not monitor and did not incur the cost of seven francs. After each round, participants were informed about the total investment and return per token in the market, their payoff from investment, monitoring revenue if they inspected others, and fine if they violated and were inspected by others. Again, a group of five subjects stayed together in one market for the eight periods.

Free mobility and Asymmetric institution (N S-treatment) The free mobility treatment provided each of the ten subjects a choice to invest either in the unregulated market A or the regulated market B. At the beginning of each round, subjects first chose the market in which to invest. After the number of people in each market was known, the threshold λ was announced in the regulated market B, and the remainder of the round proceeded as in N-treatment in the unregulated market A and as in S-treatment in the regulated market B. The threshold λ and fines in the B-market were adjusted given the number of subjects in this market. See Table 2 in the NS-instructions (part 3). After each round, participants received feedback on the total investment and return per token in each market, their payoff, and inspection results in the regulated market. Extra tokens placed by a violator who was monitored appeared next the violator’s ID and were publicly observable to all ten subjects. We did not change the labeling of the two markets across rounds and treatments in order to help coordination; and also to reflect real-world settings, where specific institutions may be associated with given localities. Market A stayed unregulated, and market B stayed regulated for the duration of the NS-treatment. This continued for 6 to 16 periods (Table 2) to make the total number of rounds equal across two designs.

Free Mobility and Voting by Ballot (V-treatment) In the voting treatment, subjects were informed that regulation in each market will be determined by the majority voting rule among the subjects in the particular 12

market. In the case of a tie, the regime was chosen randomly by the computer. Subjects first decided which market to select from two markets labeled “A” and “B”. Second, after having observed the number of people in the market, they voted for sanctions or no sanctions in their chosen market. Once the regime was decided by voting, either sanctions or no sanctions treatments were implemented. After each period, subjects in both markets received feedback on the number of participants in each market, voting outcome in each market, total investment in each market, return per token in each market, their payoff and violation level in each market if sanctions were the outcome. This treatment continued for eight to fourteen periods. The theoretical predictions for each treatment, given the parameter values, are displayed in Table 1.

5

Results

There are in total 12 sessions, four in the USA and eight in Mongolia. The summary of the experimental sessions is provided in Table 2. We analyze the data from each site, US and Mongolia, both separately and pooled together. Below we present results for the pooled data. Occasionally, we compare both sites data when differences exist. Table 3 provides summary statistics for the pooled data and each separate site. In section 5.1, we analyze the data on aggregate level. We evaluate performance of institutions by group use, individual use, efficiency, population split with free mobility, equilibrium profits in two localities, monitoring under sanctions, and voting outcome. In section 5.2, we assess data in terms of individual behavior and see how individual actions vary with institutions.

5.1

Group behavior

We use the classical Nash equilibrium prediction in one locality as a benchmark and compare it with the experimental results where no sanctions were introduced. We employ Wilcoxon signed ranks test using session averages as unites of observations for matched-pairs to compare actual average values with predicted values within each treatment. Our within-subject design allows to use the same test to compare results across treatments. Therefore, we report p-values for the Wilcoxon signed ranks test, two-tailed unless indicated otherwise. Result 1 (Resource use with no regulation). With no sanctions, the resource extraction is above the socially optimal level. Efficiency of the resource use is low 13

and no different from the Nash equilibrium prediction. The mean group use and the mean individual use were slightly below the predicted Nash equilibrium level. Support. Row N in Table 3 shows that the overall average group appropriation for the twelve experiments was 280.7, as compared to social optimum of 180, however below the Nash level of 300 at a 5 percent significance level (p=0.0186). Groups were able to absorb only 61.5 percent of the surplus available in the market, which was not significantly different from the Nash equilibrium prediction of 55.6 percent (p=0.0597). With sanctions the subgame perfect Nash model predicts no overharvesting. We use the terms “regulation” and “sanctions” interchangeably. Result 2 (Effect of sanctions). With sanctions total appropriation drops dramatically and group efficiency improves substantially as compared to no regulation. Average of the group resource use was slightly above the Nash equilibrium predictions. Support.) As Table 3, row S shows average group use dropped from 280.7 per locality in the unregulated N-treatment to 193.3 with sanctions. It was slightly above the social optimum of 180 in the data pooled across subject pools (p=0.0121). In terms of efficiency, overall surplus from resource use reached on average 94 percent and it is below the predicted value, 96.8, at one percent significance level (p= 0.0037). However, there is a huge difference between the surplus absorbed in the no sanctions and the sanctions treatments (61.5 vs. 94 percent, which differ at any significance level). Examination of two institutions in one locality produced contrasting outcomes. With no regulation the resource is overused, while sanctions prevents over-extraction. Therefore, we confirm that, as in Casari and Plott (2003), sanctioning system largely resolves the overharvesting problem of CPR. Now we analyze the data in the two-locality design with free mobility, where one locality (labeled A) was unregulated and the other locality (labeled B) had sanctions. Result 3 (Effect of free mobility). Under free mobility condition, the community with sanctions sustains a much higher resource use efficiency than the unregulated 14

community. Group use in the unregulated community is higher than in the regulated community, and no different from the equilibrium prediction. As predicted, more people locate in the community with sanctions. However, the efficiency in the unregulated locality did not increased as predicted. Support The locality with sanctions experienced an overall low average group use at 191.8 (Table 3, row NS). In contrast, the average group use in the unregulated locality across twelve experiments was 280.2. As predicted in terms of direction, the appropriation in the unregulated locality was significantly higher than the resource use in the regulated locality at 1 percent level (p=0.0022). The locality with sanctions had larger population than the unregulated locality (one-tailed, p=0.003). If we look at the consistency of participation in each markets, in each session there were at least six subjects who chose the same market in more than 50 percent of the time. There was no difference between early or later periods efficiency in the unregulated market. The results indicate that in this “voting with the feet” condition, institutions matter in the sense that a much higher level efficiency (93.1 percent) is obtained with sanctions than without sanctions (47.9). Equilibrium forces put the localities in a balance such that the majority of people chose the regulated market “B”, and fewer people invested in the unregulated market “A”. However, unlike the theoretical prediction (Proposition 3) we did not observe a positive externality from the regulated locality on the unregulated locality. Average efficiency of the unregulated locality in NS-treatment was in fact lower than in the localities with N-treatment (47.9 vs. 61.5 percent). We will discuss in section 5.2 how observed self-selection by behavioral types and competition among free riders in the unregulated locality drove down the efficiency. We now look at the V-treatment where people choose the regime by majority rule within the locality. Result 4 (Choice of a regime). Under V-treatment in the majority of cases, voting resulted in the sanctioning regime. When sanctions were chosen the resource use was at the efficient level as predicted; when no sanctions were chosen, the resource was over-used.

15

Support According to the theoretical prediction, subjects will vote for sanctions in 100 percent of the cases. The data shows that in 64 percent of the time, sanctions were implemented as a result of voting across sessions (Table 3, row VV). Two types of voting dynamics are emerged. In 3 out of 6 sessions (session 7, 8, and 10) there was a convergence to sanctions in both localities such that in market “B” sanctions was the only outcome in all periods, and market “A” gradually arrived to the regulated regime with the initial play of no sanctions. However, in two other sessions (6 and 9) we observed a divergence of regimes: in market “A” no regulation dominated and in market “B” sanctions prevailed. Session 5 in MG site exhibited unsettled patterns (Figures 1a-1f). In terms of dynamics across periods, the average percent of people who voted for sanctions in market “A” was far less than in market “B” in the first period (36 vs. 74). However, gradually, the percent of people voting for sanctions rose from the fourth period 6 , and by the seventh period, the average number of votes against regulation had dropped and votes for sanctions were about the same in both localities. Conditional on the regime chosen, the average group uses were no different from the predicted levels. With sanctioning system within the locality, individual use was much smaller at 36.8 tokens than with no sanctions at 63.9 tokens (Wilcoxon rank-sum test p-value=0.0039). Efficiency in the no sanctions regime was different from the sanctions regime at 1 percent significance level (p=0.0039). The observed population split closely reflected the predicted split of 5:5 across localities. Overall 75.7 percent of the data are within the 20 percent bandwidth (i.e. in the interval [4, 6]) around the predicted value. Exact split of 5:5 was observed in 31.4 percent of data. We note that the mean percent of the population who voted for the S-regime was significantly lower in market “A” than in market “B” in both sites (p=0.0069 in MG and p=0.0166 in US). The efficiency in the locality labeled “A” was lower than that in locality “B”(p=0.0464). Mean efficiency in market “B” was no different from predicted efficiency in the S-treatment (p=0.9165). This as we will see below suggests sorting of types into distinct markets. 6

Both NS-treatment and V-treatment have longer rounds which makes the design unbalanced. For the purpose of obtaining a full dynamics over longer period of rounds we present results with all data; the results that includes only 8 rounds in each treatment were no different from results that includes all periods.

16

We now look at whether, under the sanctioning regime, monitoring decisions (inspections) were in line with the prediction. Result 5 (Monitoring). Theory predicts one hundred percent of inspections such that entire population will be inspected. In contrast, partial monitoring was sufficient to induce higher level of efficiency. Support (Table 3) Monitoring: On average 74.4, 81, and 75.3 percent of population was inspected in S, NS, and V treatments respectively. Therefore, approximately three quarter of the population were inspected in the markets with sanctions. Next we test Tiebout equilibrium hypothesis in the two-localities case. Recall that the free mobility equilibrium is characterized by partition of the population where no citizen wants to move and where profits are equalized in both localities. But we find that localities with sanctions regime had higher per person profit than the unregulated locality. Also locality labeled “A” had higher per person profit as compared to locality “B”. Result 6 (Tiebout equilibrium hypothesis). In the free mobility treatments the average earnings in the unregulated locality were lower than in the regulated locality. Support It turns out that the average per capita profits in the two localities were statistically different from each other in (NS) and (V) treatments respectively (p=0.0076, pvalue=0.0464). The average profits in the unregulated locality, were below the equilibrium level (p=0.0096, p=0.0268) respectively while in the regulated locality average per person profits were no different from equilibrium level of 173 (p=0.1823, p=0.1313). Similarly, in V-treatment locality labeled “A”has lower mean profit than locality “B” (p=0.0464). Period by period data on difference in per capita profits also does not show any convergence to zero toward the end of sessions. We believe that profits were not equalized due to self-selection resulting free riders to cluster in market “A” and cooperators in market “B”. The predatory behavior of free riders caused downward shift in their profits.

17

5.2

Individual Behavior

First, we classify individual actions and see how they vary with isntitutions. We define “cooperative appropriation” as a harvesting action which is at or below the socially optimal level. On the other hand, “free riding” will be referred to an action which is above the social optimum, xopt . Result 7 (Institution effect). Individual actions changed with institutions: the cooperative actions were prevalent under the sanctioning regime while free-riding actions were prevalent under no regulation. Support As predicted, individual behavior changed consistently with treatments. This confirms that institutions critically affect individual behavior. Percent of cooperative actions was 19.3 in (N), 20.1 in (NS) and 19 in (V) treatments with no regulation, as compared to 65.5, 73.5, and 70.7 percent respectively in (S), (NS), and (V) treatments with sanctions (Table 4, Fig. 2). Results in both sites, MG and US, reveal the same pattern in individual behavior. We further classify behavioral types based on actions taken by participants in the no mobility N-treatment and S-treatment. Previous research identified various types of behavior in lab setting. Kurzban and Houser (2005), Gunnthorsdottir et al. (2007), Isaak and Walker (1988) classify subject’s cooperativeness based on the percentage of endowment (50, 30 or 33 respectively) that subjects devote to the public account in VCM. In CPR setting Casari and Plott (2003) identify spiteful, selfish and altruistic subjects with distinct use levels: heavy user, average user and low user respectively. Fischbacher et al. (2001) using strategy method identify free riders in one third of population and conditional cooperators in the half of population. We define a “cooperator” as a person who complies with the socially optimal investment level more than fifty percent of the time in both the N-treatment and S treatments. “Free riders” exceed harvesting levels in both N and S treatments in more than half of their real choices. “Institution-responsive” types free ride in N-treatment and comply with the rules in S-treatment in more than half of the cases. Result 8 (Heterogeneity). Subjects were hetergenous in their behavior. The majority of subjects were institution-responsive; however, the proportions of cooperators and free riders were also non-negligible. 18

Support (Table 5). Overall, the data shows that 12.5 percent of the population were cooperators, 33.3 percent were free riders. The majority of the subjects (51.7 percent) were institutionresponsive; their behavior depended upon the institution. From a total of 120 subjects, only three had behavior inconsistent with either of the 3 categories (Table 5) 7 . Experimental literature documented that in VCM at the initial phase subjects contributions are high and over time cooperation deteriorates (Ledyard, 1995) simply because conditional cooperators no longer play a role of a ”sucker” (Fischbacher et al., 2001). In order to keep cooperation level at the initial high level various sorting mechanism were proposed. Mechanisms such as costly mobility or right to change the group membership in VCM (Ehrhart and Keser, 1999), insurance mechanism that increases benefits of unilateral cooperation and reduces the loss of unilateral defection in the two-person prisoner’s dilemma game (Ahn et al., 2001), membership based on the past contribution of the counterpart (Page et al. 2002), sorting of types in the bargaining round based on offers in the previous dictator game’s round (Charness, 2000), auctioning of the rights to play the insured prisoner’s dilemma game (Bohnet and Kubler, 2005), matching of types based on trust score (Rigdon et al. 2007), matching of high-contributors with high contributors (Gunnthorsdottir et al. 2007), grouping in the VCM with punishment (Ones and Putterman, 2007) allows mostly partial sorting. In our two-locality treatments the presence of two alternative markets allows sorting. Interestingly, labeling of the markets such as “A” and “B” served as a coordinating device for the market choice decision. Result 9 (Sorting by type). In a majority of the cases cooperators and institutionresponsive types clustered in market “B” with sanctions and voted for sanctions, and free riders went to market “A” that had no regulation and voted for no regulation. Overall, the institution-responsive types monitored more than other types. At the same time, if sanctioning was the voting outcome in the locality labeled “A”, then free riders were more engaged in monitoring activity. From Figure 3 we can see that free riders on average chose regulation 43 percent of the time while institution-responsive types in both MG and US sites went to the regulated market 67 and 54 percent of the time respectively (Fig. 3). The NStreatments in MG site show that free riders went to the regulated market significantly 7

Composition of types stays the same with the classification of behavioral types based on the single period action. Classification was based on either 1st, 6th, or 8th period.

19

less often than institution-responsive ones (Wilcoxon-Mann-Whitney ranks sum test p=0.0004) while there was no difference in US site (p=0.619). Recall that we used the same labeling (“A” and “B”) in V-treatment as in NStreatment. This labeling helped institution-responsive types to join market “B”, expecting sanctions (S), and free riders to cluster in market “A”, expecting it to have no sanctions (N). In the voting treatments, institution-responsive types in the US site went to the B market more often than free riders (Wilcoxon-Mann-Whitney test p=0.0226). Behavioral types in MG site were no different in their actions: both insititution-responsive and free riders mostly ended up in the regulated locality. Besides the sorting of the population, we also observed an interesting phenomenon which we call “push out others”. The more opportunistic subjects in the unregulated market were harvesting huge amounts and suffered current losses in an attempt to earn more surplus the next period by persuading concurrent competitors to leave the market. There were five cases where one or two subjects extracted excessive amounts, resulting in market losses for everyone (e.g. unregulated locality in NS-treatment in the sessions 3, 4, 8, 12; V-treatment of the session 9). These losses caused the majority to leave the market. This behavior is similar to the predatory pricing behavior common in competitive industries (Tirole 1988). However, with some experience in V-treatment, free-riders realized that the“push out others” strategy was wasteful, leading to reductions in surplus. Non-compliant subjects in the unregulated locality learned that with sanctions they could earn more profit than without sanctions. This is in line with Plott and Li (2009); Brown, Kamp and Plott (2009), who find that buyer-preferred collusive equilibrium is attainable within the collusion incubator environment; if it was not reached then non-cooperative types such as maverick were responsible. Now we look at differences across types in monitoring and voting behavior. The percent of inspections done by institution-responsive subjects was higher than the percent of monitoring by cooperators at 5 percent significance level (p=0.0206). At the same time, in V-treatment, if sanctioning was the voting outcome in locality labeled “A”, then percent of monitoring in market “A” was significantly higher compared to the monitoring in the “B” market that had sanctions (90.5 vs. 68.8 , p=0.0350, Table 3, Fig. 4). Recall that market “A” was mostly occupied by free riders. Hence, in this market, free riders highly monitored others if sanctions regime was the voting outcome. This result is similar to Casari and Plott (2003) finding that spiteful subjects were more willing to engage in monitoring activity which induced users to follow rules of the community. In our free mobility experiments, due 20

to self-selection, free riders had less of a chance to have sanctions and monitoring outcome because their modal choice of institution was often no regulation. Majority of the inspections was done by institution-responsive types, about 1/5 of inspections was implemented by free riders, and the least monitoring was done by cooperators. In the voting treatment both localities could choose sanctions, however this did not happen from the beginning. There was a sorting of population according to their types which resulted in the implementation of sanctions 80 percent of the time in market “B”, and no sanctions 53 percent of the time in market “A”. In US site free riders on average voted for regulation 35 percent of the time, while cooperators and institution-responsive types voted for sanctions 66 and 70 percent of the time respectively. In US site, institution-responsive types voted for S-regime more often than free riders (Wilcoxon-Mann-Whitney test p-value=0.0351). However, in MG site, free riders and institution-responsive types voted for monitoring 54 and 69 percent of the time respectively; the difference was insignificant (Fig. 5). Therefore, behavioral type sorting was more observable in endogenous setting in US site, while in MG site sorting of types into different regimes was more evident in the exogenous institutions setting. This suggests that subjects in MG site, Vtreatment used the labeling from the NS-treatment for their decision to join markets less often than US subjects.

6

Discussion

This paper reports the results of an experiment that studied CPR problem with two localities. In the one-locality and two-locality designs, behavior is compared to point predictions for the symmetric Nash/SPNE equilibrium and the socially optimal outcome. We constructed four treatments to test the predictions of the free mobility model. In particular, with no regulation, the community overexploited it’s resources beyond the socially optimal level and efficiency was significantly lower than efficiency in the locality with sanctions. Our one-locality treatment results are in line with the Casari and Plott (2003) finding that sanctioning leads to improved efficiency of CPR use. Interesting results are derived from the treatment where one locality adopts sanctions and the other does not have regulation. As predicted, the sanctioning system keeps the resource use at an efficient level. However, the positive externality from sanctions was not observed in the unregulated locality, instead efficiency was dropped to the level below the predicted value. We believe that the clustering of free riders in the unregulated market, predatory behavior and rivalry among free riders lowered the efficiency in the unregulated market. Similarly, in the voting treatment, 21

one market was efficient because it was occupied by institution-responsive types who voted for sanctions whereas free riders clustered in the other market and voted less often for sanctions. However, results based upon the realized regime were in line with the predictions. Furthermore, we find that behavioral types differed in their voting and monitoring decisions in the following way: free riders voted for no sanctions most of the time while institution-responsive types preferred sanctions. Overall institutionresponsive types monitored more than the other behavioral types. However, if the sanctions were implemented in market “A” then percent of monitoring by free riders was highest. “Voting with the feet” and “voting with the ballot” conditions produced similar results by sorting out types into two distinct communities. However, comparison across two conditions reveals a short-term advantage of the former construct in terms of policy design. In the sanctions condition where the institution was given in advance, the welfare increased from the very beginning as compared to no sanctions. In the “voting with the ballot” condition subjects ended up with the social dilemma unresolved, or learned only after a few periods that sanctioning is the better way to sustain the resource, and sometime for policy makers, this suggests that the local participatory decision may take time to establish the regulatory institution. Our results draw several conclusions. Most importantly, we find that the sanctioning institution under the free mobility condition may survive if the target harvesting adjusts to the migration process. In our experiments with exogenous institutions, welfare increased from the beginning, while with endogenous institutions, it took time to establish the appropriate regime. Partial monitoring was sufficient to obtain high efficiency with sanctions. The presence of multiple-localities may bring about the sorting of subjects according their behavioral types such that cooperative types cluster in the community with sanctions and non-cooperative types in unregulated locality. This might be one reason why some regions are slow in success while others quickly progress. The positive externality from sanctions locality, which was predicted by free mobility equilibrium was not absorbed by neighboring locality in the exogenous setting due to clustering of types. Similarly, with endogenous setting clustering slows an adoption of sanctions in the locality with free riders whereas other locality that occupied by mostly institution-responsive types was able to achieve efficiency quickly as possible.

22

References [1] Ahn, T.-K., E. Ostrom, D. Schmidt, R. Shupp, and J. Walker. “Cooperation in PD games: fear, greed and history of play.” Public Choice 106, 2001, 137-55. [2] Andreoni, J., W. Harbaugh, and L. Vesterlund. “The carrot or the stick: Rewards, punishments and cooperation.” American Economic Review, 93, 2003, 893-902. [3] Banfield, E. C., and L. F. Banfield. The Moral Basis of a Backward Society. Glencoe, IL: The Free Press, 1958. [4] Bohnet, I., and D. Kubler. “Compensating the cooperators: is sorting in the prisoner’s dilemma possible?” Journal of Eonomic Behavior and Organization, 56 (1), 2005, 61-76. [5] Botelho, A., G. W. Harrison, L. M. Costa Pinto, and E. E. Rutstrom. “Social norms and social choice.” University of Central Florida, Working Paper, 2005. [6] Brown, A. L., H. Kamp, C. R. Plott, and H. J. Sullivan. “The Nature of collusion facilitating and collusion breaking power of simultaneous ascending price and simultaneous descending price auctions.” Economic Inquiry, 47 (3), 2009, 395-424. [7] Casari, M., and C. R. Plott. “Decentralized management of common property resources: experiments with a centuries-old institution.” Journal of Economic Behavior and Organization, 51, 2003, 217-47. [8] Casari, M. “On the design of peer punishment experiments.” Experimental Economics, 8, 2005, 107-115. [9] Charness, G. “Bargaining efficiency and screening: an experimental investigation.” Journal of Economic Behavior and Organization, 42, 2000, 285-304. [10] Deacon, Robert T., D. P. Parker, and C. Costello. “The Efficiency Gains from Coordinating Effort in a Fishery: Evidence from the Chignik Salmon Cooperative.” Working paper, 2008. [11] Deacon, R. T., D. P. Parker, and C. Costello. “Improving Efficiency by Assigning Harvest Rights to Fishery Cooperatives: Evidence from the Chignik Salmon Co-op.” Arizona Law Review, 50 (2), 2009, 479-510. 23

[12] Decker, T., A. Stiehler, and M. Strobel. “A comparison of punishment rules in repeated public good games.” Journal of Conflict Resolution, 47, 2003, 751-72. [13] Downs, A. An Economic Theory of Democracy. Harper Collins, 1957. [14] Duncan, B. “On the rationale of group decision-making.” Journal of Political Economy, 56, 1948, 23-34. [15] Ertan, A., T. Page, and L. Putterman. “Can endogenously chosen institutions mitigate the free-rider problem and reduce perverse punishment?” Brown University, Working Paper, 2005. [16] Ehrhart, K-M., and C. Keser. “Mobility and cooperation: on the run.” Working paper, 1999. [17] Falk A., E. Fehr, and U. Fischbacher. “Appropriating the commons: a theoretical explanation.” in Drama of the Commons, edited by E. Ostrom, T. Dietz, N. Dolsak, P. C. Stern, S. Stonich, and E. U. Weber. Committee on the Human Dimensions of Global Change, National Research Council, National Academy Press, 2002, 157-92. [18] Fehr, E., and S. Gachter. “Altruistic punishment in humans.” Nature 415 (10), 2002, 137-40. [19] Fischbacher, U., S. Gachter, and E. Fehr. “Are People Conditionally Cooperative? Evidence from a Public Goods Experiment.” Economics Letters , 71, 2001, 397-404. [20] Fischbacher, U. “z-Tree: Zurich Toolbox for Ready-made Economic Experiments.” Experimental Economics, 10(2), 2007, 171-78. [21] Fiorina, M. P., and C. R. Plott. “Committee decisions under majority rule: an experimental study.” American Political Science Review, 72(2), 1978, 575-98. [22] Gardner, R., M. Moore, and J. Walker. “Governing a groundwater commons: a strategic and laboratory analysis of western water law.” Economic Inquiry, 35(2), 1997, 218-34. [23] Greenberg, J., and S. Weber. “Strong Tiebout equilibrium under restricted preference domain.” Journal of Economic Theory, 38, 1986, 101-11.

24

[24] Gurerk, O., B. Irlenbusch, and B. Rockenbach. “The competitive advantage of sanctioning instiutions.” Science, 312, 2006, 108-10. [25] Gunnthorsdottir, A., D. Houser, K. McCabe, and H. Ameden. “Disposition, history, and contributions in public goods.” Journal of Economic Behavior and Organization, 62, 2007, 304-15. [26] Hardin, G. “The tragedy of the commons.” Science, 162, 1968, 1243-48. [27] Isaac, M., and J. Walker. “Group size effects in VCM provision: the voluntary contributions mechanism.” Quarterly Journal of Economics, 103, 1988, 179-99. [28] Kurzban, R., and D. Houser. “Experiments investigating cooperative types in humans: a complement to evolutionary theory and simulations.” PNAS, 102(5), 2005, 1803-07. [29] Ledyard, J. “Public goods: a survey of experimental research.” in Handbook of Experimental Economics, edited by J. Kagel, and A. Roth, Princeton University Press, Princeton, 1995, 11194. [30] Ones, U., and L. Putterman. “The Ecology of Collective Action: A Public Goods and Sanctions Experiment with Controlled Group Formation“. Journal of Economic Behavior and Organization, 62 (4), 2007, 495-521. [31] Ostrom, E. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press, New York, 1990. [32] Ostrom, E., J. Walker, and R. Gadner. “Covenants with and without a sword: self-governance is possible.” American Political Science Review, 86, 1992, 40417. [33] Page, T., L. Putterman, and B. Unel. “Voluntary association in public goods experiments: reciprocity, mimicry and efficiency.” Working Paper, Brown University, 2002. [34] Li, J., and C. R. Plott. “Tacit collusion in auctions and conditions for its facilitation and prevention: equilibrium selection in laboratory experimental markets.” Economic Inquiry, 47 (3), 2009, 425-48. [35] Rigdon, M., K. McCabe,and V. Smith. “Sustaining Cooperation in Trust Games.” Economic Journal., 117 (522), 2009, 991-1007. 25

[36] Sefton, M., R. Shupp, and J. Walker. “The effect of reward and sanctions in the provision of public goods.” University of Nottingham, Working Paper, 2002. [37] Sutter, M., S. Haigner, and M. G. Kocher. “Choosing the stick or the carrot? Endogenous institutional choice in social dilemma situations”, Working paper, 2006. [38] Schmitt, P., and J. S. Kurtis, and J. M. Walker. “Collective action with incomplete commitment: experimental evidence.” Southern Economic Journal, 66 (4), 2000, 829-54. [39] Tiebout, C. “A pure theory of local expenditure.” Journal of Political Economy, 64, 1956, 416-24. [40] Tirole, J. ‘The Theory of Industrial Organization, MIT Press, 1988. [41] Visser, M. “Welfare implications of peer punishment in unequal societies.” Goteborg University, Department of Economics, Working Papers in Economics 218, 2006. [42] Velez, M. A., J. K. Stranlund, J. J. Murthy. “What motivates common pool resource users? Experimental evidence from the field.” Journal of Economic Behavior and Organization, 70 (3), 2009, 485-97. [43] Vyrastekova, J., and D. Soest. “Centralized common-pool management and local community participation.” Land Economics, 79 (4), 2003, 500-14. [44] Walker, J., R. Gadner, and E. Ostrom. “Rent dissipation in a limited-access common-pool resource: experimental evidence.” Journal of Environmental Economics and Management, 19, 1990, 203-11. [45] Walker, J.M., R. Gardner, A. Herr, and E. Ostrom. “Collective choice in the commons: experimental results on proposed allocation rules and votes.” Economic Journal, 110, 2000, 212-34. [46] Wooders, M. H. “The Tiebout Hypothesis: Near optimality in local public good economies.” Econometrica, 48 (6), 1980, 1467-86.

26

Table 1: Equilibrium Predictions Social N ash/SP N E M ax Eqlm Ef f − cy Optimum Eqlm Surplus Surplus percent 5 X opt = 180 X = 300 Π = 1080 Π = 600 55.6 opt x = 36 x = 60 π = 120 (S) 5 X opt = 180 X = 180 Π = 1080 Π = 1045 96.8 P P opt x = 36 x = 36 = π− k π = 209 p(monitor) = 1 opt (N,S) 10 Xi = 180 X ( U )A = 288 Π = 1080 Π(U )A = 691.2 64.0 (loc.A) i = A, B X ( R)B = 180 per loc. Π(R)B = 1038 96.1 (loc.B) P P n(U )A = 4 = π− k n(R)B = 6 x(U )A = 72 x(R)B = 30 π(U )A = 172.8 π(R)B = 173 p(RB , monitor) = 1 (V) 10 Xiopt = 180 X(A) = 180 Π = 1080 ΠA = 1045 96.8 i = A, B X(B) = 180 per loc. ΠB = 1045 nA = 5 per loc. nB = 5 xA = 36 xB = 36 πA = 209 πB = 209 regimeA − sanctions regimeB − sanctions p(RA , monitor) = 1 p(RB , monitor) = 1 N- number of subjects; SPNE-subgame perfect Nash equilibrium; Eqlm-equilibrium; Eff-cy -efficiency ; (N) -no sanctions; (S)-sanctions; (NS)-No sanctions in one and Sanctions in other market; (V)-Voting; (A)-labeling for market with no sanctions in (NS)-treatment; (B)-label for the market with sanctions in (NS)-treatment; (A)-label for market 1 in (V)-treatment; (B)-label for market 2 in (V)-treatment; (U)-unregulated; (R)-regulation or sanctions; π-per capita profit; Π-surplus excluding endowments; X-total group appropriation, x-individual appropriation; k-per capita monitoring cost; p(monitor)-probability of monitoring. 27 T reat ment (N)

N

Table 2: Summary of Experimental Sessions Number of Date Session Code Site periods Design1 : N, S, N S (N ), (S), (N, S) 8, 8, 16 May 21, 2007 11 US 8, 8, 16 May 24, 2007 12 US 8, 8, 16 June 5, 2007 1 MG 8, 8, 16 June 7, 2007 2 MG 8, 8, 16 June 8, 2007 3 MG 8, 8, 16 June 9, 2007 4 MG Design2 : N, S, N S, V (N ), (S), (N, S), (V ) 8, 8, 6, 8 May 22, 2007 9 US 8, 8, 6, 10 May 25, 2007 10 US 8, 8, 6, 14 June 6, 2007 5 MG 8, 8, 6, 10 June 7, 2007 6 MG 8, 8, 6, 14 June 8, 2007 7 MG 8, 8, 6, 14 June 9, 2007 8 MG US-University of Hawaii at Manoa, Honolulu, USA ; MG-Academy of Management, Ulaanbaatar, Mongolia .

28

Table 3. Mean efficiency, group use, individual use, population, and inspection.

Mean Average efficiency, %

(stdev) Treatment (N)

Predi cted 55.6

(S)

96.8

(NS) N S (VV) VA

VB Regime N

64 96.1 96.8

96.8 55.6

(VV)

Predi cted

Actual

All

MG

US

61.5

61.3

61.8

(13.7)

(17.7)

(4.6)

94*

93.6*

94.8

(3.2)

(3.7)

(2.3)

47. 9*

47

49.7

(21.1)

(25.5)

(16.2)

93.1*

93.3

92.9

(3.8)

(3.5)

(3.2)

72.8*

80.9

56.5

(15.6)

(5.5)

(18.4)

90.9

88.3

96

(10.4)

(12.3)

(2.0)

59.1

67.3

42.6

(25.5)

(5.3)

(2.6)

frequency,%

0

36

36

36

Regime S

96.8

94.2*

93.3

96.2

(41.5)

(3.7)

(1.7)

frequency,%

100

Average Group appropriation (tokens)

300 180 288 180 180

180 288

180

Average individual use (tokens) Predi cted

Actual

All

MG

US

280.7*

282.1

277.9

(21.5)

(25.2)

(13.7)

193.3*

191.5

196.8

(14.0)

(15.5)

(11.5)

280.2

276.8

286.9

(24.1)

(27.1)

(18.2)

191.8*

189.7

196.1

(13.3)

(13.7)

(13.2)

246.7*

238.9

262.3

(17.0)

(11.7)

(17.5)

199.3

205

187.5

(33.2)

(40.8)

(10.8)

276.5

270.1

288.3

(11.9)

(5.9)

(9.5)

60 36 72 30 36

36 60

36

Predi cted

Actual

All

MG

Average # of people per locality

US

56.1*

56.4

55.6

(4.3)

(5.0)

(2.7)

38.7*

38.2

39.4

(2.8)

(3.1)

(2.3)

65.4*

65.2

65.8

(8.8)

(10.9)

(3.4)

33.9*

33.4*

34.8

(2.6)

(2.8)

(2.0)

49.5*

47.3

53.9

(4.2)

(2.1)

(4.0)

40.5

42.1

32.2

(9.6)

(11.2)

(7.1)

63.9

64.7

62.2

(23.8)

(9.7)

(12.1)

191.9

193.9

188.2

36.8

36.4

37.7

(15.4)

(21.4)

(0.6)

(12.8)

(1.9)

(0.4)

Predi cted

Actual

All

MG

Percent of actions inspected, % Act Ual

US

All

~

~

~

~

~

~

~

~

~

100

~ 74.4* (16.2)

4 6 5

5

4.3*

4.3

4.4

(0.4)

(0.4)

(0.2)

5.7*

5.7

5.6

(0.4)

(0.4)

(0.2)

5

5.1

4.9

(0.4)

(0.3)

(0.7)

5

4.9

5.1

(0.4)

(0.3)

(0.7)

4.6

4.5

5

(0.7)

(0.8)

(0.6)

5

5.5*

5.7*

5.2

(0.4)

(0.3)

(0.3)

5

~ 100

~ 80.9* (14.3)

100

90.5* (7.5)

100

68.8* (18.8)

~

100

~

75.3* (13.6)

64

64

64

p-value:** N=S

0.0022

0.0117

0.0679

0.0022

0.0117

0.0679

0.0022

0.0117

0.0679

~

~

~

p-value:** N=S in NS

0.0022

0.0117

0.0679

0.0022

0.0117

0.0679

0.0022

0.0117

0.0679

0.0025

0.0140

0.0679

~

0.9156 0.0747

0.5807

0.6547

0.0679

0.6547

p-value:** VA=VB

0.0464

0.1441

0.1797

0.0464

0.1441

0.1797

0.1730

0.4652

0.1797

p-value:** VN=VS

0.0277

0.0679

0.1797

0.0277

0.0679

0.1797

0.0277

0.0679

0.1797

~

0.0350 ~

*-different from predicted value at 5 percent level in Wilcoxon signed ranks two-sided test statistics that compare actual averages with predicted value; standard deviations are in parentheses; **Wilcoxon matched-pairs signed ranks test, two sided; N- no sanctions; S-sanctions; NS-no sanctions in one and sanctions in other market; VV-voting in two markets, VA-voting in A market, VB-voting in B market, VN- N-regime in voting, VS- S-regime in voting.

Regime

Action

Table 4. Percent of cooperative actions by institution (N vs. S) MG site US site Treatment N S All N N 19.1 ~ 19.1 19.6 S ~ 65.8 65.8 ~ NS 23.6 76.1 53.8 13.4 By regime, V All 18.9 69.8 53.3 19.1 loc. A 21.4 55 40.1 19 loc. B 14 81 66.1 20 Total 6.2 47.1 53.3 6.7 % of time each regime was chosen in V-treatment All 36 64 100 36 loc. A 48 52 100 67 loc. B 25 75 100 6

S ~ 65 68.3

All 19.6 65 44.1

All pooled data N S 19.3 ~ ~ 65.5 20.1 73.5

All 19.3 65.5 50.6

73.5 62.1 77.3 47.8

54.4 33.3 74.2 54.4

19 20.6 14.5 6.3

70.7 56.1 79.9 47.3

53.6 38.9 68.3 53.6

64 33 94

100 100 100

36 53 20

64 47 80

100 100 100

Cooperative action refers to harvesting at or below social optimal level; loc.-locality, N-no sanctions, S-sanctions, NS-sanctions in one and no sanctions in other locality, V-voting treatment

Table 5. Composition of individuals by behavioral types, % Treatment (N), (S)

Behavioral Types

All, %

# subjects

Unconditional cooperators Free riders Conditional cooperators Irrational

12.5 33.3 51.7 2.5

15 40 62 3 120

Cooperators: cooperate >50% in both (N) & (S) Free riders: free ride >50% in both (N) & (S) Rational: free ride >50% in (N) & cooperate >50% in (S) Irrational: cooperate >50% in (N) & free ride >50% in (S)

MG site All, % subject, # 13 34 51 3

10 27 41 2 80

1

2

3

4

1 3 6 0

2 0 8 0

2 3 4 1

3 3 3 1

Session 5 6 2 6 2 0

0 2 8 0

7

8

0 6 4 0

0 4 6 0

US site All, % subject, # 13 33 53 3

5 13 21 1 40

9

10

11

12

4 2 4 0

1 3 6 0

0 4 6 0

0 4 5 1

Figure 1a. Voting outcom e: regim e 1-sanctions, regim e 2-no santions, (MG site) session 5.

Figure 1b. Voting outcom e: regim e 1-sanctions, regim e 2-no sanctions, (MG site) session 6. 3

2

A market

1

B market

Regime

Regime

3

2

A market

1

B market

0

0 1

2

3

4

5

6

7

8

9 10 11 12 13 14

1

2

3

4

5

Period

8

9

10

Figure 1d. Voting outcom e: regim e 1-sanctions, regim e 2-no sanctions, (MG site) session 8. 3

2

A market

1

B market

Regime

3 Regime

7

Period

Figure 1c. Voting outcom e: regim e 1-sanctions, regim e 2-no sanctions, (MG site) session 7.

2

A market

1

B market

0

0 1

2

3

4

5

6

7

8

1

9 10 11 12 13 14

2

3

4

5

6

7

8

9 10 11 12 13 14

Period

Period

Figure 1e. Voting outcom e: regim e 1-sanctions, regim e 2-no sanctions, (US site) session 9.

Figure 1f. Voting outcom e: regim e 1-sanctions, regim e 2-no sanctions, (US site) session 10.

3

3

2

A market

1

B market

Regime

Regime

6

2

A market

1

B market

0

0 1

2

3

4

5

Period

6

7

8

1

2

3

4

5

6

Period

7

8

9

10

Figure 2. Percent of cooperative harvesting actions by treatment 100 Percent

80

76

66 65

70 74

68

60 40 20

MG site 24

19 20

13

19 19

0 N

S

N in NS S in NS N in V S in V Treatment Cooperative action refers the harvesting at or below social optimal use, results are robust with the threshold as a cut-point

US site

Figure 3. Mean percent of times each behavioral type went to regulated market 90

100 Percent

80 60

67 52

54

43

65 43

40

69

68 68 38

32

20 0 MG-S in NS

US- S in NS

MG-S in V

US-S in V

Treatment Cooperator

Institution-responsive type

Free rider

Figure 4. Inspections by behavioral types and treatment 100

Percent

80 60 40 20

24

32 20

26 13

29

22

31

26 28

7

0 MG-NS

US-NS

MG-V

US-V

Treatment Cooperator

Institution-responsive type

Free rider

16

Percent

Figure 5. Mean percent of times each behavioral types vote for Sregime in V-treatment

100 80 60 40 20 0

69

66

54

70 35

14 MG

US Treatment

Cooperator

Institution-responsive type

Free rider

INSTRUCTIONS (N)

instructions for no sanctions condition in a single market

This is an experiment in decision-making. The instructions are simple and if you follow the instructions carefully and make good decisions you may earn a considerable amount of money. Your earnings and $5 show up fee will be paid in CASH in private at the end of the experiment. Conversion rate is $______ per 1 experimental franc. You are NOT allowed to communicate with any other participant. From this point onwards, you will be referred to by your participant number. Your ID number is at the left top corner of the screen. You will make money by INVESTING tokens in a market that will give you a cash return for your tokens (See example #1 below). The experiment in which you are participating is comprised of several parts, each having a sequence of periods. In each period you will be asked to make an investment in a MARKET. PART 1: INVESTING In this part of the experiment, you will be in a market four other participants. What happens in your market has no effect on the people in the other market, and visa versa. You will not be told which people are in your market, but you will be in the market with the same four other people during Part 1 of the experiment. In each period you can invest a number of tokens between 0 and 500 in the market. All other participants can also invest up to 500 tokens and so the total group investment is at most 2500 (= 500 times 5 people). For every token you invest, you will be charged 2.5 francs, and you will be collecting your return in experimental francs from the market. The return from the market depends on the number of tokens you invest as well as the amount all others in the group invest. The total group investment determines the gross group return (See attached table A) and you will receive a fraction of it according to your personal investment level. Your earning depends on total group investment, gross group return, number of tokens you invest, cost, and endowment in a following way

Your Earnings=

Gross Group Re turn x Your Investment - Cost + Endowment Total Investment Your Share of Gross Group Return

The example below explains the computation in detail. You will make your decision before knowing other people’s investment decisions in that period. You are not to reveal your investment decision to anyone. Other participants’ decisions are private to you; however you will be informed about the total group investment and your profit at the end of each period. EXAMPLE #1 Suppose you invest 20 tokens in the market and the rest of participants invest in total 200 tokens. The cost of your tokens is 50 francs: Cost of your Investment= 2.5 francs per token x 20 tokens = 50 francs In this example, the total group investment is 220 tokens (20 tokens you invest plus 200 tokens by others). The corresponding gross group return is 1577 francs, as shown in the table A attached. The first column of the table lists the total group investment and the second column gives you the corresponding gross group return. The last column indicates return per token invested and obtained by dividing gross group return (in 2nd column) to the total group investment (in 1st column).

You will receive a share of the gross group return. For your share of gross, you can multiply your personal investment level by the “Return per token invested” column of the table A, that is 20 *7.17 =143.4. Your net return is 93.4 francs (your share of gross, 143.4, minus the cost of the tokens, 50). Finally, your earning for each period is the sum of your net return and endowment that will be given to you in advance. The period endowment is a constant amount and does not depend on the investment decisions. To find your period earning add values in columns (8) and (9) of the following table. To sum up, in this example your period earnings in francs are given by (assume you are ID #5): ID #

Number of token invested (your investme -nt)

Total Group Invest ment

Gross Group Return

Return per token

(table A)

(table A)

(1)

(2)

(3)

(4)

5

20

220

1577

Your Share of Gross

(5)=(4)/(3)

(6)=(2)x(5)

7.17

143.4

Cost

Net return

(7)=2.5 x (2)

(8)=(6)-(7)

-50

93.4

Endo wme nt

Period Earning

(9)

(10)=(8)+(9)

+10

103.4

The gross group return is graphed below.

Gross Group Return (francs)

Gross Group Return on Market 1600 1450 1300 1150 1000 850 700 550 400 250 100 -50 -200 0 -350

48

96

14

4

2 19

24

0

28

8

33

6

4 38

43

2

48

0

52

8

6 57

62

4

67

2

72

0

8 76

81

6

86

4

91

2

0 96

Total Group Invetsment (tokens)

Notice that the gross group return on the market can be negative if the total group investment is sufficiently large. For instance, if each person invests 110 tokens, the total group investment is 550 tokens and the gross group return is –200 francs. When considering the cost of the tokens, each person has to pay 315 francs. In each period you may record your investment and earnings information in the record sheets provided. This investment decision will continue for a number of periods. Your earnings for each period will be added to find your cumulative earnings for this part of the experiment. If you have any questions concerning the instructions feel free to raise your hand and an instructor will assist you. Please, go through the review question in the next page and fill in the blank lines with the values you think are correct.

PART 1_N

PRACTICE ID ______

REVIEW Consider the following investment decisions: ID# 1 2 3 4 5 Total group investment tokens 40 40 40 60 60 240 Suppose you are person #1. To compute your net return on the market, take your investment of 40 tokens and multiply it by 6.5 (Return per token invested, second column of the table A) = 260 francs is your share of gross. Your net return is 160 francs (your share of gross = 260 minus the cost of tokens = 100). Now, you go on and fill in the blank lines for ID#2 and ID#5. ID #

Number of token invested

(1)

(2)

1 2 5

40 40 60

Total Group Invest ment

Gross Group Return

Return per token

(table A)

(table A)

(3)

(4)

240

1560

(5)=(4)/(3)

6.5

Your Share of Gross (6)=(2)x(5)

260

Cost

Net return

(7)=2.5 x (2)

(8)=(6)-(7)

-100 -

160

Endo wme nt

Period Earning

(9)

(10)=(8)+(9)

+10 +10 +10

170

Please, raise your hand if you have any questions and an instructor will assist you.

At the beginning we will run a practice-period experiment to get familiar with the rules. It will NOT count towards your earnings.

ARE THERE ANY QUESTIONS?

Table A. GROSS GROUP RETURN Total group investment (1)

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390 400 410 420 430 440

Gross Group Return (2)

0 142 277 405 527 642 750 852 947 1035 1117 1192 1260 1322 1377 1425 1467 1502 1530 1552 1567 1575 1577 1572 1560 1542 1517 1485 1447 1402 1350 1292 1227 1155 1077 992 900 802 697 585 467 342 210 72 -50

Return per token invested (3)=(2)/( 1)

0.00 14.17 13.83 13.50 13.17 12.83 12.50 12.17 11.83 11.50 11.17 10.83 10.50 10.17 9.83 9.50 9.17 8.83 8.50 8.17 7.83 7.50 7.17 6.83 6.50 6.17 5.83 5.50 5.17 4.83 4.50 4.17 3.83 3.50 3.17 2.83 2.50 2.17 1.83 1.50 1.17 0.83 0.50 0.17 -0.11

Total group investment (1)

450 460 470 480 490 500 510 520 530 540 550 560 570 580 590 600 610 620 630 640 650 660 670 680 690 700 710 720 730 740 750 760 770 780 790 800 810 820 830 840 850 860 870 880 890

Gross Group Return (2)

-116 -152 -173 -185 -192 -195 -197 -198 -199 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200

Return per token invested (3)=(2)/( 1)

-0.26 -0.33 -0.37 -0.39 -0.39 -0.39 -0.39 -0.38 -0.38 -0.37 -0.36 -0.36 -0.35 -0.34 -0.34 -0.33 -0.33 -0.32 -0.32 -0.31 -0.31 -0.30 -0.30 -0.29 -0.29 -0.29 -0.28 -0.28 -0.27 -0.27 -0.27 -0.26 -0.26 -0.26 -0.25 -0.25 -0.25 -0.24 -0.24 -0.24 -0.24 -0.23 -0.23 -0.23 -0.22

Total group investment (1)

900 910 920 930 940 950 960 970 980 990 1000

Gross Group Return (2)

-200 -200 -200 -200 -200 -200 -200 -200 -200 -200 -200

Return per token invested (3)=(2)/( 1)

-0.22 -0.22 -0.22 -0.22 -0.21 -0.21 -0.21 -0.21 -0.20 -0.20 -0.20

1. Find out the actual group investment 2. Look at the average return of each token 3. Multiply the average return by the number of tokens you have invested

INSTRUCTIONS (S) instructions for the sanctions condition in one market PART 2: In this part of the experiment you will make your money by INVESTING them in the market that will give you cash return for your tokens as in a previous part of the experiment. Also you may earn money by MONITORING other people’s decisions and eventually getting some revenues from the inspections. Each period consists of two stages: Investment and Monitoring. The investment procedure will be exactly the same as in the PART 1 of the experiment. After the investment stage you will have chance to earn additional revenue from monitoring others decisions. MONITORING After the total group investment is revealed, you will have the chance to impose a payment to the people that invested more than 34 tokens in the market. Notice that if the total group investment is more than 170 tokens (that is 34 tokens times 5 people), at least one person invested more then 34 tokens. You don’t know the individual investments of the other people, but you can ask to uncover them by paying 7 francs for every person you ask to inspect (inspection fee). If the person inspected invested more than 34 tokens, she pays 4.8 franc for every extra token. You get this money (inspection revenue) and everybody will know the investment level of the person inspected. You make the monitoring decision when you know the total group investment, but before knowing other people’s monitoring decisions. An identification number will be assigned to every person to maintain anonymity and it must be considered strictly confidential. EXAMPLE #2 Suppose the total group investment is 280 tokens and your investment is 40 tokens (you are ID #1). Before the monitoring you know that 110 extra tokens were invested (=280 -170) but you don’t know who invested them. Well, you did part of the job with 6 tokens (=40 – 34), but there are other four people around that invested 104 extra tokens. Suppose you ask to inspect person #2 and she has invested 64 tokens. You pay an inspection fee of 7 francs and get an inspection revenue of 144 francs (= (64 – 34) * 4.8) where 4.8 is the fine per each extra token invested. After your inspection, everybody will know that pers#2 has invested 64 tokens, but your identity will not be revealed. In sum, besides the period earnings from investment already explained above, your earnings will be affected by your and other people’s monitoring decisions as follows: Monito EndowPeriod Cost of Monito Fine if Your Total ID Number Inspecte ring ment Earning tokens -ring Share Group # of token d cost revenu of Invest invested e Gross ment (1)

(2)

(3)

1 2

40 64

280

(4)

(5)=2.5 x (2)

206.8 330.8

-100 -160

(6)=[(2)34]x4.8

(7)

(8)

+144 ~

~ -144

-7 ~

(9)

+10 +10

(10)=(4)(5)+(6)-(7)(8)+(9)

253.8 36.8

If you (ID#1) get inspected, then you will pay for every token above 34 and this fine would be (4034) x 4.8=28.8. If two or more people ask to inspect the same person, only one inspection will be executed. A person will be randomly selected by computer and she will pay the inspection fee and get the eventual inspection revenue. The other inspectors will be treated as if they did not ask to inspect that person. ARE THERE ANY QUESTIONS?

INSTRUCTIONS (N-S) PART 3:

instructions for the two-asymmetric market free mobility condition

In this part of experiment you will make your money by INVESTING them in the market that will give you cash return for your tokens. There are two markets: MARKET A and MARKET B. You can only invest into one market in any period. If you are investing in MARKET B you may earn money by both investing and monitoring other people’s decisions. MARKET A has investment and no monitoring. In each period you will be asked to make either an investment in MARKET A or an investment and a monitoring in MARKET B decision. There are NINE other participants in this experiment who will be asked to choose which market to invest, and then to make investment decisions. All participants who choose market B will also make monitoring decisions. Note that maximum number of tokens free of charge and monitoring fine per each extra token invested change with a number of people in a MARKET B, however monitoring cost stays the same. See the attached Table B with the fine and monitoring cost in MARKET B. EXAMPLE #3 Suppose you (ID#1) and two other people invest in MARKET A with a total group investment of 200 and seven other persons decide to invest in MARKET B with a total group investment of 200 tokens. ID#

1

2

3

Marker A Market B

40

100

60

4

5

6

7

8

9

10

40

40

25

26

24

22

23

Total group investment 200 200

Therefore, return per token invested is the same, 7.83 francs, in both MARKETS since the total group investment is identical. Suppose ID#5 inspects ID#4, therefore receives monitoring revenue, 82.24=(40-24)x5.14 (see Table B for fines and maximum fine-free number of tokens). At the same time ID #5 pays a fine for extra 16 tokens=40-24, because someone else inspects her as well. After the market period everyone will know that ID #5 and ID #4 have invested 16 extra tokens each. ARE THERE ANY QUESTIONS?

Table B. Monitoring fine and cost in MARKET B Number of people in a Market 1 2 3 4 5 6 7 8 9 10

Fine-free Maximum number of tokens per person 0 87 58 43 34 28 24 21 18 16

Fine per extra token

Monitoring cost per person

0.00 3.00 4.00 4.50 4.80 5.00 5.14 5.25 5.33 5.40

0 7 7 7 7 7 7 7 7 7

INSTRUCTIONS (VV) PART 4:

instructions for the two-market voting condition VOTING

In this part of the experiment, you will first decide which MARKET to choose, A or B. After market choices, you will be asked to select, or vote for, exactly one of two alternatives, NO MONITORING and MONITORING. All participants in your market will make their decisions at the same time, without knowing the choices of others. The alternative that collects most of the votes will be the outcome in your market. If there is a tie between two alternatives, both alternatives get the same number of votes, and then the outcome will be chosen randomly. The decisions made in the other MARKET have no effect on your earnings. EXAMPLE #1 Suppose the number of participants in your chosen market is 6 and the number of votes for NO MONITORING is 4 and number of votes for MONITORING is 2, then NO MONITORING will be the outcome. EXAMPLE #2 Suppose the number of participants in your chosen market is 6 and the number of votes for NO MONITORING is 3 and number of votes for MONITORING is 3. This is a tie. Then the outcome will be chosen randomly between NO MONITORING and MONITORING. Please, go through the review question and fill in the blank lines with the values you think are correct. PRACTICE ID ______ REVIEW Consider the following voting decisions: Suppose MARKET A has 4 people and MARKET B has 6 persons. Given the numbers of alternatives listed below find the outcome of voting in each market. Number of votes for NO MONITORING in MARKET A: 3 Number of votes for MONITORING in MARKET A: 1 The outcome in market A:____________________________ Number of votes for NO MONITORING in MARKET B: 3 Number of votes for MONITORING in MARKET B: 3 The outcome in market B: ____________________________ Please, raise your hand if you have any questions and an instructor will assist you.

ARE THERE ANY QUESTIONS?