Social capital in the commons:

does it really make a difference? ∗

Nuria Oses-Eraso† and Montserrat Viladrich-Grau�

†Unidad de Economıa Ambiental

Universidad del Paıs VascoAvenida Lehendakari Agirre 83

Bilbao 48015, Spaine-mail address : nuria.oses@ehu.es

�Department of Business Administration (AEGERN)Universitat de Lleida

ETSEA, Rovira Roure 191Lleida 25198, Spain

e-mail address :montse.viladrich@aegern.udl.es

February 13, 2008

∗The authors gratefully acknowledge the financial support received from the Government of Navarraunder its research support program “Ayudas a la investigacion 2002-2003” and CICYT grant numberSEJ2006-11510. We also acknowledge the helpful comments of Jorge Alcalde-Unzu, Miguel A. Ballesterand Frederic Udina. The experiments described in this paper were implemented while the first authorwas at the Universidad Publica de Navarra.

1

Social capital in the commons:does it really make a difference?

Abstract

We implement an experimental study of the appropriation of common poolresources in a dynamic setting with social capital where extinction is a crediblethreat. Resource availability depends on the initial environmental characteristicsof the common resource and on human-induced resource depletion due to users’appropriation patterns. Our results show that participants link appropriation andsocial capital. Appropriation determines the use of the resources for social capi-tal and the distribution of social capital is used as an incentive to maintain lowappropriation levels and to delay the destruction of the common resource. In par-ticular, we show that the presence of social capital enhances resource preservationand limits the appropriation strategies of the agents sharing the resource. Fur-thermore, we find that agents distribute social capital in shares inversely relatedto appropriation.

Keywords: Common property resource, social capital, appropriation, bestowals,concern for resource preservation, early extinction, scarcity.

2

1 Introduction

Many common property resources are exploited by local communities where group in-

teraction plays an important role. The mere concept of common property presupposes

the existence of a well defined group of users who are free to interact strategically with

each other (Baland and Platteau, 1996). This interaction starts with joint resource

exploitation and is complemented, in many cases, with other types of social links that

flourish among community members and that give rise to social capital (Ostrom, 2000,

Adler and Kwon, 2002). The joint exploitation of the resource has a powerful impact on

the development of social interaction mechanisms, which, in turn, influence appropria-

tion strategies. In these communities social capital has deep historical roots and these

linkages among community members give the collectivity cohesiveness and facilitates

the pursuit of collective goals.

The social relationships linked to resource appropriation take a variety of forms,

ranging from the development of social norms and management rules, through the es-

tablishment of different types of punishments and rewards, to the exchange of favors

between community neighbors that conform a social structure where, in addition to

the benefits obtained in the exploitation of the common resource, an individual can

obtain social capital based on the relations with others and on their attitude towards

the common resource (see the examples in Ostrom, 1990, Ostrom et al., 1994, Baland

and Platteau, 1996). In this setting, the ability to obtain social capital is directly linked

with the behavior of each agent in the exploitation of the common resource. To posses

3

social capital, a person must be related to others and they are those others, not him-

self, who are the actual source of his or her capital. In the present paper, we focus on

analyzing the role of social capital in these type of settings where a local community

jointly exploits a natural resource. In particular, we analyze the role of social capital in

the exploitation of a common property resource from an experimental point of view.

In these communities the provision of social capital is linked with appropriation of

the common resource. Each member of the community exploits the resource and gets

benefits from this activity. His appropriation decisions affect other members of the

community. Likewise, he is affected by the appropriation of the other members of the

community. Also, in the commons, the ability to obtain social capital is related with

the behavior of each agent in the exploitation of the common resource. The members

of these communities are engaged in two linked activities, a productive activity (the

joint exploitation of the common resource) and a social activity, and therefore interact

in both stages. In our experiment, to mimic this setting, all subjects are both, a source

and a receptor of social capital. We distinguish among the possessors or recipients of

the social capital, the sources or donors of social capital and the resources themselves.

Each community member faces a two stages game. In the first stage, he jointly exploit

a common resource and makes appropriation decisions. In the second stage, he is a

receptor and a donor of social capital. We refer to the first stage as the appropriation

stage and to the second as the social capital stage. We focus on analyzing the link

between these two stages. In particular, on one side, we study whether the availability

4

of resources for social capital is able to affect the receptors’ appropriation strategies, and

on the other, whether the assignation patterns of the donors of social capital is affected

by receptors’ appropriation strategies.

Furthermore, in these communities, as Ostrom (2000) says, bestowing social capital

can take many forms such as social approval, prestige or status, but in all of them social

capital differs from physical capital in that it does not wear out with use but rather

with disuse. Resources are only transformed into effective social capital if they are

distributed among community members. Thus, the resources for social capital that are

not distributed are lost. Moreover, in many collective action problems, the provision of

social capital often appears to be relatively cheap (Fershtman and Weiss, 1998, Ostrom

et al., 1994). Therefore, in our experimental design social capital can be bestowed

cost-free in the interaction stage and can be lost if it is not distributed.

Further, in this study we depart from the traditional time-independent common

pool resource (CPR) games used in most experiments, where the trade-offs in a CPR

setting are described without any explicit mention of resource stock size.1 Ours is

a time-dependent setting where today’s appropriation determines tomorrow’s resource

availability. The resource stock is exogenously given only in the first round; thereafter,

it is endogenously determined by the agents’ appropriation strategies. The maximum

duration of the game is also common knowledge and the scarcity conditions are fully

1The traditional CPR experiment is based on the long-run sustainable yield function that charac-terizes renewable resources and relates exploitation effort to sustainable yield. A complete descriptionof the intricacies of the original biological model can be found in any manual of natural resource eco-nomics, see Tietenberg (2006) as an example. Also, for a full description of the traditional CPR gamesee Ostrom et al., (1994).

5

defined. Moreover, extinction is a credible threat. The abundance or scarcity of a

resource stock may be due either to initial environmental conditions or to common

resource users’ behavior.

Although the experimental design introduces great incentives for preservation, our

results show that social capital is able to improve the efficiency in the exploitation of

the common resource although a substantial part of the resources for social capital are

not distributed by the donors and they are, therefore, lost. Thus, the possibility of en-

joying social capital limits receptor’s appropriation. The proportion of participants that

manage to preserve their resource increases with social capital, irrespective of the initial

resource stock size. Additionally, we conclude that scarcity is relevant in determining

appropriation behavior patterns as we observe that the scarcer the common resource

becomes, the lower the appropriation. Moreover, appropriation history matters, agents

follow the appropriation trend and do not counteract the appropriation strategies of

previous rounds, often they are unable to offset the reduction in stock size. Further-

more, we show that the assignation patterns of social capital are affected by the receptor

previous appropriation decisions. The distribution of the social capital is directly liked

with the behavior of the subjects in the appropriation stage. The more cooperative

subjects are both the main donors and the main receptors of the social capital. More

precisely, we focus on identifying the characteristics associated to the main receptors

of the social capital. Finally, we analyze the criteria followed by the donors of social

capital.

6

The paper is organized as follows. In section 2 we present the two stages game

and the corresponding benchmarks. Section 3 establishes the experimental design and

procedure. The main results are presented in section 4. Finally, the main conclusions

are in section 5.

2 Social structure: the game

The social structure is represented by a two stages game. In the first stage (appropriation

stage), subjects interact in the appropriation of the common resource. In the second

stage (social capital stage), they interact in the distribution of social capital.

2.1 Appropriation stage

This stage describes the commons dilemma, a situation where individual behavior,

driven by the maximization of the individual payoffs, leads to actions that are socially

suboptimal. Let a community of n agents share a common pool resource of F0 points.

Each community member has an endowment of e points, which she can invest in project

A or in project B. Investment in project A, (that we call appropriation), reports an indi-

vidual marginal benefit w but causes a decline of c points in the common pool resource.

On the other hand, investment in project B has an individual marginal benefit α and

causes the common resource to grow by g points. Let xit be the appropriation of subject

i in round t; then the appropriation payoff that she obtains in such round is:

πAit = wxit + α(e − xit) (1)

7

After the investment decisions of round t, the remaining resource, Ft = Ft−1 −

c∑n

i=1 xit + g∑n

i=1(e − xit), represents the common pool resource that the subjects

share for the next round of investments. The investment decisions can be taken over a

maximum of T rounds, the time horizon, as long as the common pool resource maintains

a positive value. Further, when the common pool resource takes a negative or zero value,

there exists early extinction and the investment rounds conclude. In addition, if the last

round T is reached, the payoff in this round is increased by dividing the remaining

resource, FT equally among the n community members.

We assume that(

w − cn

)

>(

α + g

n

)

and (w−c) < (α+g) to describe the traditional

social dilemma associated with common pool resources; the individual net benefit from

project A,(

w − cn

)

, is greater than the individual net benefit from project B,(

α + g

n

)

,

while the aggregate net benefit from project A, (w− c), is lower than the aggregate net

benefit from project B, (α + g).

2.2 Social capital stage

Now assume that each subject j observes, after each appropriation round, the appro-

priation decisions of the other members of her community. Then, she receives a new

endowment, d, and can voluntarily decide whether she wants to distribute all or part of

this endowment among the other community members. This new endowment represent

the resources for social capital. We define yijt as the bestowal made by subject j to

agent i, for any i 6= j, and such that 0 ≤ yijt ≤ d. The total amount of social capital

8

that can be bestowed by subject j is Djt =∑

i yijt such that 0 ≤ Djt ≤ d. The resources

for social capital that are not distributed are simply lost, Ljt = d − Djt.2 Observe that

the distribution of social capital does not provide any direct material benefits for the

donor, the direct benefits are for those that receive the points, the recipients.3

As a recipient of social capital, each subject i can receive individual bestowals, yijt,

from any donor j 6= i of her community. The total social capital that she receives is

Rit =∑

j yijt and 0 ≤ Rit ≤ (n − 1)d. This social capital represents the payoff that

subject i can obtain in the second stage of round t, that is, the social capital payoff.

Thus, in any round t, πSCit = Rit =

∑

j yijt.

The total payoff of subject i in round t, πit, is the sum of her payoffs in each stage:4

πit = πAit + πSC

it = wxit + α(e − xit) +∑

j

yijt (2)

Therefore, suppose that the total number of rounds played by community members

is t∗, where t∗ ≤ T ; then the total payoff obtained by subject i is equal to:

Πi =

∑Tt=1 πit + FT

nif t∗ = T and FT > 0

∑t∗

t=1 πit otherwise

(3)

In this setting, early extinction occurs whenever t∗ < T .

2A similar social structure with two stages and two endowments has been used by other authorsin other social dilemmas but the new endowment is never lost. For example, in Sefton et al. (2007),subjects participating in a public goods game also receive a new endowment after contributing. Thisnew endowment can be used to reward or punish but can also be put in a private account, that is, thenew endowment is never lost.

3Recall that, to possess social capital a person must be related to others and it’s those others andnot himself who are the actual source of his social capital (Portes, 1998), therefore, a subject can notobtain direct benefit from her new endowment.

4Observe that πSCit ∈ [0, (n − 1)d] while πA

it ∈ [αe, we] and recall that in round T the payoff functionis incremented by an equal share in the remaining resource, if any.

9

2.3 The benchmarks

We solve the game by backward induction to determine the best individual strategy,

xNiT . Subject i maximizes the total payoff in round T , ΠiT .

ΠiT = wxiT + α(e − xiT ) +FT

n+ RiT (4)

where FT = FT−1 + gne − (c + g)∑n

i=1 xiT . The FOC of this maximizing problem are:

∂ΠiT

∂xiT

=(

w −c

n

)

−(

α −c

n

)

+∂RiT

∂xiT

(5)

where the solution depends on the relationship between the appropriation of agent i

during period T , xiT , and the bestowals received in period T , RiT , represented by ∂RiT

∂xiT

.

We can make different assumptions about the distribution of these resources for social

capital. First, if the distribution of social capital does not depend on the appropriation

decisions, ∂RiT

∂xiT

= 0, then the best individual strategy is full-appropriation in round T ,

xNiT = e, as we have assumed that the individual net benefit of project A,

(

w − cn

)

,

is greater than the individual net benefit from project B,(

α + g

n

)

. Second, if the

distribution of social capital negatively depends on the appropriation decisions, ∂RiT

∂xiT

<

0, then full-appropriation is still the best strategy if(

w − cn

)

−(

α − g

n

)

>∣

∣

∣

∂RiT

∂xiT

∣

∣

∣ for

all xiT . On the contrary, non-appropriation is the best individual strategy, xNiT = 0,

whenever(

w − cn

)

−(

α − g

n

)

<∣

∣

∣

∂RiT

∂xiT

∣

∣

∣ for all xiT . Also, if(

w − cn

)

−(

α − g

n

)

=∣

∣

∣

∂RiT

∂xiT

∣

∣

∣

for an appropriation level x∗

iT such that 0 < x∗

iT < e, the best individual strategy is

xNiT = x∗

iT . Finally, it is also worth mentioning that the best individual strategy in

10

round T will be full-appropriation whenever the distribution of social capital positively

depends on the appropriation decisions, ∂RiT

∂xiT

> 0.

Now consider the investment decision in the next to last round, (T −1). In the cases

where xNiT = e, the maximizing problem in (T − 1) has the following form.

Πi(T−1) =

[

wxi(T−1) + α(e − xi(T−1)) + Ri(T−1)

]

+

+[

we +F(T−1)−cne

n+ RiT

]

if F(T−1) > cne

[

wxi(T−1) + α(e − xi(T−1)) + Ri(T−1)

]

+

+[

we + Ri(T−1)

]

if 0 < F(T−1) ≤ cne

(6)

The maximizing problem is similar to the one that appears in a simple game where

there is only an appropriation stage (see Oses-Eraso et al. (2008) for a complete descrip-

tion of the game without social capital). Subject i has incentives to appropriate less

than the total endowment if this guarantees a minimal level of resource stock and future

payoffs different from zero. Therefore, whenever ∂Rit

∂xit

= 0, ∂Rit

∂xit

> 0 or ∂Rit

∂xit

< 0 with

(

w − cn

)

−(

α + g

n

)

>∣

∣

∣

∂Rit

∂xit

∣

∣

∣ the resource will be exhausted up to his minimal survival

level in the shortest possible time and thereafter subjects will face a coordination game

till the last round of the game.

When non-appropriation is the best strategy in round T , that is, when(

w − cn

)

−

(

α + g

n

)

<∣

∣

∣

∂RiT

∂xiT

∣

∣

∣, in period (T − 1) subject i maximizes the following payoff function,

Πi(T−1) =[

wxi(T−1) + α(e − xi(T−1)) + Ri(T−1)

]

+

[

αe +F(T−1) + gne

n+ RiT

]

(7)

The FOC are the ones in T (see equation 5) and something similar happens in other

rounds of the game. Therefore, if(

w − cn

)

−(

α + g

n

)

<∣

∣

∣

∂Rit

∂xit

∣

∣

∣, the best individual

11

strategy is non-appropriation in every round.

When the best strategy in round T is xNiT = x∗

iT and if we assume that ∂Rit∂xit

is the

same in any round t, an appropriation xNit = x∗

it in every round is the best strategy

whenever x∗

it ≤ xLit where xL

it is the limit strategy, the maximum average appropriation

that avoids early extinction (see Oses-Eraso, 2008). However, if x∗

it > xLit, subjects will

face a coordination game in some round of the game.

Observe that the solution of the game depends on the way the resources for social

capital are distributed. Nothing can be said about this distribution unless we consider

the expectations of the agents.5

From an aggregate point of view, however, the best strategy is to invest the whole

endowment in project B in the appropriation stage (non-appropriation, xEit = 0) and to

distribute the whole new endowment in the social capital stage. This maximizes the

aggregate benefit of the community. Nothing can be said, however, about how the new

endowment should be distributed.

5In this context, if subjects act as sources and bestow social capital on other community members,the open question is why they do so. Recent economic models consider people to be inequity averse,that is, to show a dislike for outcomes that they perceive to be inequitable (Fehr and Schmidt,1999,Falk el al., 2000). Inequity aversion may be behind the bestowal of social capital. Nevertheless, lookingat some regularities found in economic experiments, Falk et al. (2001) indicate that subjects sanctionnot only because they want to reduce distributional inequities but also to punish unfair intentions.Furthermore, rewards can be used as a selective incentive to influence the behavior of others (Oliver,1980).

12

3 Experimental design and procedure

The experiment was conducted using the program z-tree (Fischbacher, 1999) in the

computer rooms of the Universidad Publica de Navarra. All the participants were un-

dergraduate students from various disciplines, all of them from this University.

The game was presented to the participants as a two-stages game. During the

experiment, we referred to the stages as the first and the second stage. The first stage

is the appropriation stage and it is characterized by the investment options open to the

subjects. Subjects have an endowment of e = 20 experimental points and know that

investment in project A (appropriation) reports an individual marginal benefit of w = 2

points causing a decrease in the common pool resource of c = 3 points, while investment

in project B has an individual marginal benefit of α = 1 points and causes an increase

of g = 0.75 points. The time horizon is T = 20 rounds. We implement two different

treatments, an scarcity treatment where F0 = 1200 and an abundance treatment where

F0 = 2400.

The participants know that, once appropriation decisions have been taken, they have

to play a second stage. This second stage is the social capital stage. At the beginning of

this stage, the computer displays information about the individual appropriations and

the appropriation payoffs of each community member.6 Each group member receives a

new endowment of d = 20 points that she can distribute between the other members

6In order to prevent reputation-building, the first column on the screen displays the player’s ownappropriation level, while the individual appropriations of the other community members appear ran-domly in the three remaining columns.

13

of her community. Participants are aware that any undistributed points will simply be

lost. Once the distribution has been completed, the participants receive information

about the earnings in each stage, the total earnings of the round and the remaining

resource stock.

To study the effect of social capital on appropriation, we also implemented an abun-

dance treatment and an scarcity treatment without the social capital stage.7 Therefore,

we analyzed the results of four different treatments (see table 1).

Without Social Capital With Social Capital

Abundance 8 groups (32 subjects) 8 groups (32 subjects)Scarcity 8 groups (32 subjects) 8 groups (32 subjects)

Table 1: Treatments

Observe that in all the treatments, there is a risk of early extinction. The minimum

duration of the game is 10 rounds in the abundance treatment and just 5 rounds in the

scarcity treatment, with and without social capital stage.8

We run two different sessions per treatment. Each subject participates in only one

treatment. Subjects were randomly divided into matching groups of four members and

the composition of the groups remained unchanged for the whole session. Moreover,

the actual composition of the groups was not known to the participants. On arrival,

7See Oses-Eraso et al (2008) for a complete analysis of the treatments without social capital stage.8The best individual strategy is full-appropriation (xit = 20) till round 9 in the abundance treat-

ments and till round 4 in the scarcity treatments when there is no social capital or in situations where theexpectations on social capital do not compensate for a reduction on appropriation. Non-appropriation(xit = 0) is the best social strategy in the appropriation stage in any of the treatments. We canalso consider a the sustainable strategy, xit = 4, and the limit strategy, xit = 12 in the abundancetreatments and xit = 8 in the scarcity treatments.

14

the subjects had to read the instructions9 where it was emphasized that there would be

no communication with other participants during the experiment. Furthermore, after

reading the instructions, the participants had to answer a comprehension text. Before

the experiment began, all answers where check to make sure that the participants had

understood the structure of the game.

Subjects were told that at the end of the experiment the points would be exchanged

for cash, at a prespecified exchange rate. Each session lasted around 1 hour and the

average earnings per subject amounted to about 15 Euros.

4 Results

We divide the results into two different subsections. First, we consider the effect of

the resources for social capital on the appropriation strategies, that is, we focus our

attention on the decisions taken in the appropriation stage. And second, we analyze

the distribution of the social capital, that is, the sources and the recipients of the social

capital.

4.1 Results on appropriation

We compare agent’s behavior in the appropriation stage under four different treatments:

abundance without social capital, abundance with social capital, scarcity without social

capital and scarcity with social capital. We center the study on the differences due

9The instructions can be found in Appendix A.

15

to the presence of resources for social capital.10 The big box in figure 1 contains four

panels, A, B, C and D, that represent the average appropriation time series for each of

the communities that participated in each treatment. In addition, we have depicted in

each panel the treatment average appropriation with a solid line. On the right hand side

of figure 1, panels E and F compare treatment average appropriation with and without

resources for social capital.

Figure 1: Appropriation: social capital vs. no-social capital.

It is easy to observe that average appropriation is lower when there are resources for

social capital. Moreover, a consequence of the higher appropriation levels when there

are no resources for social capital is that half of the communities fail to reach the final

round of the game, that is, haft of the communities are not able to avoid early extinction.

This happens no matter the initial quantity of the common resource (in panels A and

10The effect of diferent levels of initial resource stock were studied in a previous paper (Oses-Eraso,2008).

16

C of figure 1 early extinction is marked with a big dot). On the contrary, when there

are resources for social capital, all the communities manage to reach the final round of

the game.11 The following observations summarize these facts.

Observation 4.1 The availability of resources for social capital reduces appropriation

for any level of initial resource scarcity.

Observation 4.2 The rate of resource survival increases when there are resources for

social capital for any level of initial resource scarcity.

For the statistical analysis of the differences on average appropriation, we use the

test suggested by Cuevas et al. (2004) that compares time series.12 The differences

in average appropriation are significant with abundant (p = 0.002) and with scarce

(p = 0.0001) initial resources. Therefore, the presence of resources for social capital

significantly reduces average appropriation in common property resources.13 Although

average appropriation is still above the best social strategy (zero appropriation) when

there are resources for social capital, the observed decline in average appropriation

increases efficiency in the exploitation of the common resources.14

11It is worth mentioning that, in any round t, early extinction has been avoided and the preservationof the common resource is certain if the remaining common resource is such that Ft > cne(T −t). Whenthere are resources for social capital, all the communities are able to secure preservation by round 16in the abundance treatment and by round 17 in the scarcity one (note, that, given the design of ourtreatments, the game can not become one of certainty before round 8 in the abundance treatment orbefore round 12 in the scarcity treatment).

12We use the mean-padding approach to complete the time series of the communities that do notreach period T (see Oses-Eraso et al., 2008).

13According to the benchmarks, this significant reduction in average appropriation can be a signalof a negative relation between appropriation and distribution of social capital. We analyze this pointin the following subsection.

14At the individual level, the efficient strategy represents a 47.64% of the individual strategies in the

17

The differences found in average appropriation when there are and when there are no

resources for social capital have a direct effect on the evolution of the common resource.

In the first situation (resources for social capital) a high quantity of the common re-

source is preserved, no matter if the resource was initially abundant or scarce.15 On the

contrary, when the are no resources for social capital, on average the common resource is

nearly exhausted by period T = 20, even when it was initially abundant. The evolution

of average common resource is depicted in figure 2.

Figure 2: Evolution of the common resource

To further study the effect of social capital on individual appropriation we estimate

panel data regression model where our dependent variable is the appropriation of agent

i during round t, xit.16 We introduce several independent variables that we grouped

abundance treatment with social capital and a 55.31% of the individual strategies in the scarcity one.The choice of this strategy is smaller when there is no social capital (25% and 13%).

15Half of the communities with resources for social capital reach round T with a common resourcegreater than the initial one. The common resource stock in round T in the abundance treatmentwith resources for social capital is in the interval [532.50, 3540.00] and in the scarcity treatment withresources for social capital is in the interval [476.25, 2036.25]. Recall that the initial common resourcewas 2400 in the abundance treatment and 1200 in the scarcity one.

16As we said in section 3, we have several communities in our data set but, for simplicity we do notinclude a subindex to differentiate them since there are no interactions among members of differentcommunities, a subject i can only belong to a community and interact with its members.

18

in three types: i) the variables that capture the evolution of the resource stock, ii) the

social capital variables, and iii) the socioeconomic variables.

As the investment rounds progress, the scarcity faced by each community is de-

termined by the appropriation strategies followed by their community members. In

each round, the resource stock can increase (if (c + g)∑n

i=1 xit < gne) or decrease (if

(c + g)∑n

i=1 xit > gne). In order to test whether this variation of the resource stock

influences players’ strategies, we include two independent variables to capture the ef-

fect of this endogenous resource variation.17 The first captures the positive variation

in the resource stock, if any, which is represented by PVit. It measures the increment

in the resource stock size in a community during the previous round and is defined as

PVit = gne − (c + g)∑n

i=1 xit−1 whenever (c + g)∑n

i=1 xit−1 < gne and 0 otherwise.18 If

the regression coefficient of this variable is positive and significant it will imply that an

increase in the resource stock size is followed by an increase in appropriation decisions.

That is, an increase in PVit is followed by an increase in xit. Agents’ behavior will

tend to erode the increase in resource stock size in the previous round. If the estimated

regression coefficient of PVit turns out to be negative and significant, however, it will

imply that an increase in the resource stock size is followed by a reduction in appropri-

ation decisions taken in the next period. Agents will tend to preserve and sustain the

increasing trend of the resource stock.

Similarly, we define NVit, as the variable that captures the negative variation of

17When scarcity is generated by the community members’ decisions we call it endogenous.18All agents i in the same community and in the same round t face the same PVit.

19

the resource stock, if any. It measures the decrease in the resource stock size during

the previous round. It is equal to NVit = |gne − (c + g)∑n

i=1 xit−1| whenever (c +

g)∑n

i=1 xit−1 > gne and 0 otherwise. If the regression coefficient of this variable is

positive and significant, it will imply that a reduction in the resource stock size is

followed by an increase in the average appropriation decisions taken in the next round.

Agents’ behavior will tend to exacerbate the previous reduction in the resource stock

size. If, however, it turns out to be negative and significant, it will imply that a reduction

in the resource stock size will be offset by a reduction in appropriation decisions in the

following round.

Initial scarcity can also affect appropriation strategies. Therefore, to identify and

isolate the effect of initial scarcity conditions, we introduce as explanatory variables the

dummy variable DAi, which takes a value of 1 if agent i belongs to a community with

an initial resource of 2400 units (abundance treatment) and zero otherwise.19

Moreover, to confirm whether appropriation strategies are affected by the actual

scarcity level, we introduce the scarcity index, SIit. The initial resource stock, the

evolution of the resource stock and the time horizon together determine the actual

scarcity faced by subject i in a community in any round t. All else being equal, a given

resource level would be considered more abundant the fewer rounds were left to be

played. In order to measure this absolute scarcity we construct an scarcity index (SIit),

that we define as, SIit =F(t−1)

T−(t−1)where F(t−1), represents the remaining resource stock at

19We drop subindex t because these variables are constant through time.

20

the end of the previous round in the community, and T −(t−1) the number of rounds to

go (including t) before reaching the time horizon T . Furthermore, the scarcity index is

an easy-to-calculate indicator of the abundance of the resource. In each round, a player

has information about the remaining resource stock (Ft−1), the actual round (t) and the

time horizon (T ). The index is a straightforward way to measure the resource stock that

can be consumed in each round. These two characteristics of the game, the actual level of

stock and the appropriation horizon, fully define the absolute scarcity level faced by any

community in any round t.20 Two individuals facing the same SIit face the same absolute

scarcity but may be enjoying different levels of resource stock, have a different number

of remaining rounds to play and their communities can belong to different treatments.

Therefore, if SIit is significant in explaining an agent’s appropriation behavior, we could

conclude that agent’s behavior depend on the actual measurement of scarcity, and not

only on how it is reached, since it could be reach through different behavioral strategies.

This index does not capture whether the scarcity is improving or deteriorating - it only

captures the actual level. Therefore, if SIit is non-significant and this lack of significance

is accompanied by significant coefficients on PVit or NVit, we could conclude that the

evolution of the resource is more relevant than the actual scarcity level in determining

appropriation attitudes.21

Second, we introduce the social capital variable. A dummy variable SCit captures

the presence of this resources taking value 1 whenever there are resources for social

20Note that each agent i in the same community faces the same scarcity index in round t.21For a more detailed explanation of these variables see Oses-Eraso et al., 2008.

21

capital and value zero otherwise. Giving the previous results, we expect the coefficient

of this variable to be negative and significant, showing that individual appropriation

decreases in the presence of resources for social capital. We also consider the interaction

between social capital and resource variation introducing SCit × PVit and SCit × NVit.

The third set of variables represent the information about the gender of the subjects

(Gi) and about their academic education (Ei). Finally, we include a set of time dummies

in order to investigate the effect of time on appropriation decisions (T t). The results of

this regression analysis are shown in table 2.

The results obtained in our preliminary regressions (in Table 2) show that the pre-

sence of resources for social capital, SCit, has a negative and significant effect on in-

dividual appropriation. It could be interpreted that agents in the appropriation stage

“expect” to receive a large amount of social capital if they restrain their appropriation

strategies. The negative sign of PVit suggests, as we said earlier, that an increase in

resource stock size is followed by a reduction in appropriation decisions. Individuals

tend to preserve and sustain the increasing resource stock trend. Agents with low ap-

propriation levels do not switch behavior. A similar interpretation can also be made

with respect to the positive sign of the NVit coefficient. A decrease in resource stock

size causes an increase in average appropriation levels. Agents’ behavior tends to ex-

acerbate the previous reduction in resource stock size. Again, the effect is similar but

in the opposite direction; individuals with high appropriation levels find it difficult to

switch behavior. The endogenous variation determined by the appropriation pattern

22

R1 R2

Constant 7.4976 (0.9944) 7.4315 (1.0094)DAi 1.4568 (0.6740) 1.4727 (0.6958)NVit 0.0371 (0.0046) 0.0382 (0.0054)PVit -0.0027 (0.0112) -0.0764 (0.0156)SIit 0.0008 (0.0009) 0.0009 (0.0009)Gi -0.8538 (1.0519) -0.9697 (1.0595)Ei 0.8779 (0.9632) 0.9038 (0.9607)SCit -3.3214 (0.7090) -3.1480 (0.7869)SCit × NVit -0.0089 (0.0100)SCit × PVit 0.0753 (0.0186)T2 -2.0192 (0.4003) -1.8123 (0.4145)T3 -1.0076 (0.7170) -0.7944 (0.7439)T4 -1.2861 (0.5429) -1.1923 (0.5695)T5 -2.1393 (0.5242) -2.0835 (0.5193)T6 -2.3026 (0.5201) -2.3271 (0.5204)T7 -2.4543 (0.5229) -2.3942 (0.5229)T8 -1.6596 (0.6889) -1.5866 (0.7007)T9 -2.0016 (0.6503) -2.0037 (0.6407)T10 -1.8020 (0.7684) -1.7815 (0.7668)T11 -2.3208 (0.6846) -2.2560 (0.6800)T12 -2.2291 (0.6671) -2.2046 (0.6461)T13 -1.9092 (0.7690) -1.8676 (0.7445)T14 -2.4628 (0.7782) -2.4281 (0.7587)T15 -1.4108 (0.8955) -1.4233 (0.8844)T16 -2.4825 (0.8758) -2.4127 (0.8521)T17 -1.5189 (0.9330) -1.5192 (0.9087)T18 -1.2704 (0.9149) -1.2090 (0.8654)T19 -0.0803 (1.1167) -0.0291 (1.0855)T20 -0.8245 (1.7461) -0.7145 (1.7112)

N=2424 N=2424Note: Standard errors in parentheses

Table 2: Determinants of appropriation

has a significant effect on appropriation strategies. Agents do not counteract the appro-

priation strategies of previous rounds but follow the appropriation trend, with either an

increasing or decreasing effect on the resource stock. It has to be noted, however, that

the positive variation PVit is no longer significant when there are resources for social

capital.22

22It is interesting to point out that our data set shows that to encounter periods where there is apositive variation of the common resource is something exceptional when there are no resources forsocial capital but is much more general when there are resources for social capital.

23

The absolute scarcity, SIit, is positive but not significant in all regression equations,

that is, we cannot say that actual scarcity levels reduce appropriation. Moreover, there

are significant differences between the abundant and the scarce treatment as the coeffi-

cient of DAijt is negative and significant, that is, appropriation is smaller the lower the

initial amount of resource stock.

The negative sign of PVit and the positive sign of NVit show that agents do not coun-

teract the appropriation strategies of previous rounds but follow the appropriation trend

and therefore resource destruction is possible. High appropriation levels are followed by

further high appropriation levels. They are unable to offset the reduction in stock size.

Concern for resource preservation is therefore not enough to prevent the destruction of

common property resources, that is, to avoid the tragedy of the commons.

Finally, the pattern of signs and significance levels of all the regressions indicates that

there are no substantial differences among the appropriation decisions in the different

rounds.

4.2 Results on social capital

In this subsection, we are interested in the behavior of the participants in the social

capital stage and in the linkages between this stage and the appropriation stage. Recall

that the use of the resources for social capital is voluntary and has no direct cost or

benefit for the donor.

As we have already pointed out, each subject can be a donor and a recipient of social

24

capital. As a donor, subject j can distribute a maximum of 20 points as individual

bestowals. Therefore, yijt ∈ [0, 20] and so is the total social capital distributed by

subject j, Djt. As a recipient, subject i can receive a maximum of 60 points from the

other community members, that is, Rit ∈ [0, 60]. In figure 3, we have depicted three

histogram of frequencies for: i) the distributed social capital, Djt, (panels 3A and 3B),

ii) the individual bestowals, yijt, (panels 3C and 3D), and iii) the received social capital,

Rit, (panels 3E and 3F).23 The following observations come straightforward.

Observation 4.3 Only in a 50% of the cases, subjects distribute all the resources for

social capital. Moreover, in a 10% of the decisions, all the social capital is lost.

A large proportion of the available resources for social capital are lost. The use of the

resources for social capital is slightly above 75% in the scarcity treatment (an average

of 15.52 points with standard deviation 6.52) but slightly below 75% in the abundance

treatment (an average of 14.51 points with standard deviation 7.30), that is, a 25% of

the available social capital is lost.

In figure 4 we have depicted the average use of resources for social capital and the

average appropriation levels. We extract the following observation,

Observation 4.4 Low average appropriation levels go hand in hand with a high average

distribution of resources for social capital.

23These frequencies are calculated from the 640 decisions on the distribution of social capital takenin each treatment (32 subjects, 20 rounds).

25

Figure 3: Donors and recipients

The use of the resources for social capital is related with appropriation. Low average

appropriation levels go hand in hand with a relatively high use of the resources for social

capital. In fact, it is easy to observe that the lower average appropriation levels of the

scarcity treatment are accompanied by a slightly higher average use of the resources

for social capital. However, the differences in the distribution of social capital in the

abundance and the scarcity treatments are not significant (p = 0.526).24

24We use again the test suggested by Cuevas et al (2004).

26

Figure 4: Appropriation and social capital

In order to study the details of the relationship between appropriation and social

capital, we divide all the participants into two groups: one made up of subjects whose

appropriation is above the average appropriation of their community (relatively high

appropriation levels) and another with subjects whose appropriation is equal to or be-

low the average appropriation of their community (relatively low appropriation levels).

Moreover, as each subject can be a donor of social capital but also a recipient of social

capital, we link relative appropriation levels with the distribution of social capital and

the reception of social capital.

In figure 5A and 5B we depict the average distribution of the resources for social

capital made by relatively high appropriators and relatively low appropriators and we

compare it with the total average use. Similarly, in figure 5C and 5D we depict the

average social capital received by relatively high appropriators and relatively low ap-

propriators. A summary of this information can also be found in table 3.

27

Figure 5: Social capital: donors and recipients

Observation 4.5 (i) Participants with relatively low appropriation levels distribute

more resources for social capital than those with relatively high appropriation levels.

(ii) Participants with relatively low appropriation levels receive more social capital than

those with relatively high appropriation levels.

The differences in the use of the resources for social capital between high and low

appropriators are statistically significant. In both treatments, abundance and scarcity,

low appropriators donate more. On the other hand, the ones that receive more social

28

Table 3: Social capital: donors and recipients.

Relative Average social capital Average social capitalappropriation as donors (Djt) as recipients (Rit)

Abundance low (xit ≤ xt) 15.50 20.78(6.92) (8.41)

high (xit > xt) 12.60 3.12(7.65) (4.09)

Scarcity low (xit ≤ xt) 16.36 20.25(5.84) (6.80)

high (xit > xt) 13.58 5.13(13.58) (5.57)

Note: Numbers in parentheses denote standard errors.

capital are also the low appropriators. In fact, low appropriators receive more social

capital than they distribute. On the contrary, high appropriators receive less social

capital than they distribute. This evidence shows that the resources for social capital

are selectively bestowed on low appropriators.

Bestowals can be strategically motivated in the sense that people make gifts if they

expect to obtain future cooperation from their group. Social capital acts as a selective

incentive that influences the behavior of others so that “the potential incentive user must

consider the difference his use of the incentive will make to the behavior of others and the

difference their behavior will make to his payoffs” (Oliver, 1980).25 In our experimental

design, the use of the resources for social capital can facilitate the achievement of the

25Falk et al. (2001) describe strategically-motivated sanctions in public good experiments. Theyconsider a partner treatment and a stranger treatment to examine the relative importance of strategicand non-strategic sanctions. They point out that, in the partner treatment, sanctions may be drivenby strategic or non-strategic reasons, while, in the strangers treatment, they can only be driven bynon-strategic reasons. By comparing strangers with partners, they discriminate strategic sanctions.We have only considered a partner treatment, which complicates the study of strategically motivatedgifts.

29

last investment rounds and the distribution of some final resource. That is, whenever

there exists the possibility of early extinction, the use of the resources for social capital

can be strategically motivated by the desire to secure future cooperation and reach the

final round.26

People may also make bestowals in order to compensate for payoff differentials de-

rived from actions taken in the appropriation stage, that is, a kind of compensating

rule. This rule is related to the theoretical model proposed by Ferh and Schmidt (1999)

and Falk et al. (2000) who introduce preferences that present inequity aversion. In our

context, this distribution rule can be interpreted as a way of compensating agents who

choose low appropriation levels and contribute to the preservation of the common pool

resource as they are obtaining lower appropriation payoffs.27

A regression analysis will be helpful to identify the determinants of the bestowals.

We pool the data of the abundance and the scarcity treatments in a panel data. The

individual bestowals, yijt, are the dependent variable. According to the results shown so

26Spagnolo (1999) points out that the ‘defense game’ generated by a war or by the presence of acommon enemy (early extinction in out case) tends to become linked to other social and productiverelations in the community. This way cooperation becomes sustainable in all linked relations. Thedisappearance of the common enemy brings the end of the defense game and leads to a lost of socialcapital.

27At the end of each session we asked the participants to briefly describe their behavior in the rewardstage. Here are some of the answers (originally in Spanish): “I gave a lot of points to those who investedzero in project A and zero points to those who invested something in the same project. I wanted them torealize that they would not only benefit the shared pool but also receive an individual payoff in the formof rewards in the second stage”; “As the investments appeared in random order and I didn’t know whosethey were, I aimed my decisions on trying to even up the distribution. I gave less points to those withhigher payoffs and tried to benefit those with lower payoffs”;“I rewarded the group interest and punishedindividual interests”; “If someone invested more or less like I did, I give him the 20 points”; “My goalwas to increase the common pool by always investing in B. Nevertheless, when my gain expectations(18 points) were not fulfilled, I invested in A”. Although the words reward and punishment were notmentioned by the experimenters, many participants used them in their descriptions.

30

far, different factors can influence the distribution of the bestowals, mainly the behavior

of the donor in the appropriation stage and the behavior of the receivers in the appro-

priation stage. In order to capture the effect of these factors, we introduce different

independent variables.

The behavior of the donor in the appropriation stage is captured by his/her appropri-

ation, xit. We call this variable donor’s appropriation. If higher appropriators distribute

less social capital, we will get a negative coefficient whereas if higher appropriators dis-

tribute more social capital, the coefficient will be positive. However, high appropriation

is a relative term. Therefore, we introduce two variables, donor’s negative deviation

from average and donor’s positive deviation from average. The first one is defined as

the absolute value of the difference between the donor’s appropriation and the average

appropriation of his/her community, when the former is lower that the latter. Other-

wise, this variable takes value zero. Observe that a positive (negative) coefficient for this

variable will imply that low appropriators donate more (less). Similarly, we define the

variable donor’s positive deviation from average as the difference between the donor’s

appropriation and the average appropriation of his/ her community when the former

is higher that the latter. In any other case, the variable takes value zero. Therefore,

a positive (negative) coefficient will imply that high appropriators donate more (less).

According to the evidence presented so far we expect a positive sign for donor’s negative

deviation from average and a negative sign for donor’s positive deviation from average.

The behavior of the recipients of the bestowals is described by two variables: re-

31

ceiver’s negative deviation from donor and receiver’s positive deviation from donor, that

compare the receiver appropriation with that of the donor. We expect a positive sign for

the first variable and a negative sign for the second one, that is, we expect the donors to

distribute higher bestowals to those that are below their appropriation and lower ones

to those whose appropriation is above theirs.

Two dummy variables measure the effect of scarcity conditions on the distribution of

the social capital. First, the dummy variable DA takes a value of 1 when the observation

belongs to the abundance treatment while it takes a zero value when the observation

belong to the scarcity treatment. The other value is early extinction and tries to capture

whether there exists the possibility of early extinction. This dummy variable takes a

zero value when there exists a possibility of early extinction and a value of 1 when there

is no such possibility. In other words, in the first rounds, this dummy variable is zero

for every subject, whatever the treatment. But as time goes by, the dummy variable

can take a value of 1 for some subjects as the game becomes one of certainty in which

the possibility of early extinction no longer exists. Therefore, the dummy variable is not

constant through time. Recall that the coefficient of this variable will have implications

for strategically motivated bestowals. We also include two dummy variables to represent

donor’s characteristics: gender and undergraduate major in economics.

The results of the regressions are shown in table 4. Starting with the behavior of the

donor in the appropriation stage, we can observe that the donor’s appropriation itself

is not significant but his relative appropriation is. In fact, the positive sign of donor’s

32

Constant 6.1244 (0.2927)Donor’s ap. -0.0440 (0.0504)Donor’s neg. dev. from av. 0.7023 (0.0841)Donor’s pos. dev. from av. -0.3948 (0.0910)Receiver’s neg. dev from donor 0.2667 (0.0547)Receiver’s pos. dev. from donor -0.6313 (0.0592)DA 0.3207 (0.3710)Early Extinction -0.2095 (0.1977)Donor’s gender 0.5222 (0.2722)Donor’s economy -0.9547 (0.2602)

N = 3840R2 = 0.4372

Note: standard errors in parenthesis

Table 4: Determinant of individual bestowlas

negative deviation from average and the negative sign of donor’s positive deviation from

average confirm that low appropriators donate more that high appropriators. More-

over, the lower the appropriation of those that are below group average, the higher the

bestowals that they distribute. Similar, the higher the appropriation of those that are

above group average, the lower the bestowals that they distribute.

The behavior of the recipients of the bestowals in the appropriation stage has also a

significant effect on the distribution of bestowals. Those whose appropriation is below

that of the donor receive higher bestowals the lower is their appropriation and those

whose appropriation is above that of the donor receive lower bestowals the higher is

their appropriation.

Finally, the variables that measure the scarcity conditions are not significant indi-

cating that the bestowals are neither affected by initial scarcity not by the possibility of

early extinction. Therefore, the main determinants of the distribution of social capital

is the behavior of the subjects, both donors and receivers, in the appropriation stage.

33

5 Conclusions

The results obtained in this dynamic setting capture the role of social capital and show

that it can play an interesting role in local communities that jointly exploit a natural

resource. The presence of social capital can improve the efficiency in the appropriation

of the common resource enhancing resource preservation. This paper provides experi-

mental evidence to support this result. In our paper it is not costly to bestow social

capial, however, we find that only 75% of these social resources are used selectively

to enhance low appropriation and achieve a more efficient exploitation of the common

resource, the rest is lost. We can conclude that the resources for social capital reduce

appropriation levels and contribute positively to resource preservation.

Also our results also capture the role of abundance and scarcity in the appropria-

tion strategies of subjects. They highlight that scarcity, in general, limits appropriation.

However, this restriction in appropriation strategies is not enough to avoid the depletion

of the common property resources if there is no social capital. Moreover, we observe that

resources that are initially more abundant do not have a greater survival rate. The level

of initial resource scarcity is important in determining initial appropriation strategies,

particularly by inducing more caution in appropriation strategies than resource abun-

dance. Furthermore, in this setting, the level of actual scarcity is determined by both

the initial resource scarcity and the agents’ behavior. That is, the level of scarcity is a

combination of environmentally-induced scarcity and human-induced scarcity. Although

disentangling the effects of both factors in appropriation strategies is a difficult task,

34

a deeper analysis of the results shows that the restriction on appropriation is greater

when the resource was initially scarcer.

35

References

[1] Adler, P.S.; Kwon, S-W. (2002): Social capital: prospects for a new concept,Academy of Management Review, 27(1), pp.17-40.

[2] Baland, J.M.; Platteau, J.P. (1996): Halting degradation of natural resources. Isthere a role for rural communities?, Oxford University Press.

[3] Cuevas, A.; Febrero, M.; Fraiman, R. (2004): An anova test for functional data,Computational Statistics & Data Analysis, 47, pp.111-122.

[4] Falk, A.; Fehr, E; Fischbacher, U. (2000): Appropriating the Commons: A Theo-retical Explanation, Institute for Empirical Research in Economics. U. Of Zurich;Working Paper no.55.

[5] Falk, A.; Fehr, E; Fischbacher, U. (2001): Driving forces of informal sanctions,Institute for Empirical Research in Economics, University of Zurich, Working PaperNo.59.

[6] Fehr, E., Gachter, S. (2000): Cooperation and punishment in public goods experi-ments, American Economic Review, vol.90, no.4, pp.980-994.

[7] Fehr, E.; Schmidt, K.M. (1999): A theory of fairness, competition, and cooperation,Quarterly Journal of Economics, 114(3), pp.817-868.

[8] Fershtman, C.; Weiss, Y. (1998): Social rewards, externalities and stable prefer-ences, Journal of Public Economics, vol.70, pp.53-73.

[9] Fischbacher, U. (1999): z-Tree: Zurich Toolbox for Readymade Economic Exper-iments, Reference Manual and Tutorial, Institute for Empirical Research in Eco-nomics, University of Zurich.

[10] Knack, S.; Keefer, P. (1997): Does social capital have an economic payoff? Across-country investigation, Quarterly Journal of Economics, 1251-1288.

[11] Oliver, P. (1980): Rewards and punishments as selective incentives for collectiveaction: theoretical investigations, American Journal of Sociology, vol.85, No.6,pp.1356-1375.

[12] Oses-Eraso, N.; Udina, F.; Viladrich-Grau, M. (2008): Environmental versushuman-induced scarcity: do they trigger the same response?, Environmental andResource Economics, forthcoming.

[13] Oses-Eraso, N.; Viladrich-Grau, M. (2007): On the sustainability of common prop-erty resources, Journal of Environmental Economics and Management, 53, pp.393-410.

36

[14] Ostrom, E. (1990): Governing the commons. The evolution of institutions for col-lective action, Cambridge University Press.

[15] Ostrom, E. (2000): Social capital: a fad or a fundamental concept?, in Socialcapital: a multifaceted perspective, Dasgupta, P. and Serageldin, I. (Eds), WorldBank, Washington, DC.

[16] Ostrom, E.; Gardner, R.; Walker, J. (1994): Rules, games and common-pool re-sources, University of Michigan Press.

[17] Portes, A. (1998): Social capital: its origins and applications in modern sociology,Annual Review of Sociology, 24, pp.1-24.

[18] Sefton, M., Shupp, R., Walker, J. (2007): The effects of rewards and sanctions inprovision of public goods, Economic Inquiry, 45(4), pp.671-690.

[19] Spagnolo, G. (1999): Social relations and cooperation in organizations, Journal ofEconomic Behavior and Organization 38:1-25.

[20] Tietenberg, T. (2006): Environmental and resource economics, Addison-Wesley.

37

Appendix

A Instructions (originally in Spanish)

We present the instructions for the abundance repeated game. We omit the instructions forthe other treatment as they are quite similar.

Welcome to the experiment

During the experiment you are not allowed to communicate with the other participants.

Should you have any questions please ask us. We will answer your question personally.

This experiment studies decision making in an economic environment. In these instructionsyou will find information about the decisions you can take and about the consequences of yourdecisions.

Depending on your decisions you can earn money that you will receive in cash at the endof the experiment. During the experiment, we will not speak of Euros but rather of points.At the end of the experiment, the total amount of points you have earned will be convertedto Euros at the following rate:

90 points = 1 Euro

The experiment

The experiment is divided into 20 periods. In each period you have to make an invest-ment.

You are in a group of 4 persons, that is, you are in a group with three other participants.The members of your group are the same for the 20 periods although you will not know whois in your group. The earnings of your investment decisions will depend on your own decisionsas well as on the decisions of the rest of your group.

Each of the 20 rounds has two stages. In the first stage you have to take an investmentdecision. In the second stage, you receive information about the investment decisions of theother three members of your group and you have the opportunity to increase their earningsby assigning them points. Below you will find detailed information about each of these twostages.

First stage

At the beginning of the first period, your group receives a pool of 2400 points. Yourinvestment decisions can affect this pool. If the pool achieves a zero or a negative value, thegame is finished. If you reach the last period, period 20, and there is a positive quantity inthe pool, it will be divided equally among the 4 members of the group.

38

Furthermore, at the beginning of each period you receive 20 points. We call these pointsyour endowment. You have to decide how many of these points you want to invest ina project A and how many you want to put in a project B. Investment in project A plusinvestment in project B must sum 20.

In the screen, you will have to insert the points you put in project A. You can put anyinteger number between 0 and 20. The rest of your endowment points (20-investment in A)are automatically put in project B.

1. Income from project A: for each point that you invest in project A, you get 2 points. Thisinvestment also decreases the pool of your group in 3 points.

1 point in A = 2 points for you

1 point in A = 3 points less in the pool

2. Income from project B: for each point that you invest in project B, you get 1 point. Thisinvestment also increases the pool of your group in 0.75 points.

1 point in B = 1 point for you

1 point in B = 0.75 points more in the pool

Your income at the end of the first stage depend on these investment decisions: it is thesum of the payoff you get from project A and the payoff you get from project B. That is,

Income at the end of the first stage:

2 × investment in A + 1 × investment in B

The income of each participant is calculated in the same way.

At the end of the first stage, the pool varies depending on the investments of the wholegroup on project A and project B.

Change in the pool at the end of the first stage

0.75 × investment in B − 3 × investment in A

Information at the end of the first stage. At the end of the first stage, you will obtain thefollowing information in the screen:

• Your investment in project A.

• The average investment of your group in project A.

• The change in the pool: positive (increment) or negative (reduction).

• Quantity still available in the pool.

• Your income of the first stage

39

Once you have seen the information press the key continue. Then, the second stage begins.

Second stage

At the beginning of the second stage, information will appear on screen about the invest-ment decisions of the four group members and the corresponding payoffs. You then get 20points that you can use to increase the payoffs of the other three group members but not yourown.

Distribution of pointsYour investment in project A and your income appeared in blue in the first column. The

investments and the payoffs of the other three members of your group appear in the followingcolumns in a random order. Each column summarizes the information about one group mem-ber. At the end of each column you will find a blank were you have to write the points youwish to assign to that person. If you do not wish to modify someone’s payoffs, you can assignhim zero points. Each point assigned increases the payoff of that person by one point.

Remember that you have only 20 points to distribute. The sum of the points that youassign to the members of your group can be no greater than 20. You are not obliged to useall 20 available points.

Once you have distributed the points, press Total Assigned Points. The total number ofpoints you have assigned will appear on the screen. If this is your final decision, press OK toconfirm. Otherwise, you can modify your distribution.

Your income in the second stageYour income in the second stage depends on the points that the other members of your

group have assigned to you. Each point you receive is a point you get.

Your income from the roundYour income from the round is the sum of the income from the first stage and the income

from the second stage.

Information at the end of the second stageAt the end of the second stage, the following information will appear on screen:

• Your income for the first stage

• The points received in the second stage

• Your total income for the round (first stage + second stage)

At the beginning of each new round, you receive a new endowment of 20 points and theopportunity to invest in projects A and B. In the second stage, you receive 20 points thatyou can distribute among the members of your group. Remember that if the pool reaches anegative or zero value, the game is finished.

40

Your total income at the end of the experiment is the sum of the income you have obtainedin each of the rounds played, plus the equal division of the pool if your group reach period 20.

Total income if the game don’t reach period 20

Income of the periods played

Total income if the game reach period 20

Income of the 20 periods played + remaining pool4

Comprehension questions. The following examples will help you to understand the exper-iment. There are not relevant for your final earnings.

1. Not you, not other members of your group invest in project A.- Increase in the pool points- Decrease in the pool points- Change in the pool points- Your income from project A points- Your income from project B points- Your total income in the first stage points

2. Each group member invest 20 points in project A (including you).

(same questions)

3. Total group investment in A is 48 points. 5 of these points have been invested by you.- Average investment of the group in A points- Increase in the pool points- Decrease in the pool points- Change in the pool points- Your income from project A points- Your income from project B points- Your total income in the first stage points

4. Total group investment in A is 48 points. 17 of these points have been invested by you.

(same questions)

41