Journal of Public Economics 119 (2014) 19

Contents lists available at ScienceDirect

Journal of Public Economics

journal homepage: www.elsevier.com/locate/jpube

Resource scarcity and antisocial behavior

Sebastian Prediger a, Bjrn Vollan b,, Benedikt Herrmann c
a
b
c

GIGA German Institute of Global and Area Studies, Neuer Jungfernstieg 21, 20354 Hamburg, Germany
University of Innsbruck, Universittsstrae 15, 6020 Innsbruck, Austria
University of Nottingham, School of Economics, University Park, Nottingham, NG7 2RD

a r t i c l e

i n f o

Article history:
Received 9 December 2013
Received in revised form 25 April 2014
Accepted 23 July 2014
Available online 4 August 2014
Keywords:
Antisocial behavior
Conict
Resource scarcity
Competition
Lab-in-the-eld experiment
Namibia
Pastoralists
Commons

a b s t r a c t
This paper examines whether exposure to persistent resource scarcity on the commons affects pastoralists' readiness to engage in antisocial behavior towards their fellow commons users. The region under study is divided into
two areas according to exogenous variations in biomass production stemming from geological peculiarities. We
conducted a joy-of-destruction game with pastoralists from both areas and provide evidence for a positive relationship between antisocial behavior and long-term exposure to scarcity. Antisocial behavior among villagers
occurs twice as often in an area where resources are scarcer and competitive pressure is higher. Our results
remain robust to the inclusion of various controls, including beliefs, socio-demographics, absolute and relative income, and within-group inequality. If one interprets decreasing another participant's payoff below one's own as a
tendency towards conict behavior, our results suggest a higher risk of conict among resource users in areas of
greater scarcity.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Scarcity is a fundamental principle in economics. In a world of scarce
resources, human needs exceed the means, and scarcity conditions
human behavior. Scarcity may be conducive to social welfare as it provides incentives to allocate resources efciently. Yet, scarcity may also
encourage people to engage in antisocial activities that are detrimental
to overall welfare, such as theft or conict. So far, the relationship between resource scarcity and conict behavior has been mainly explored
at the macro level, with a strong focus on violent intra- and interstate
conict due to climate shocks (e.g. Burke et al., 2009; Homer-Dixon,
1999; Hsiang et al., 2013; Miguel et al., 2004; Zhang et al., 2007) and
to a lesser extent (and with mixed evidence) due to different resource
endowments (Brunnschweiler and Bulte, 2009; Collier and Hoefer,
1998). At the most basic level however, conict and other consequences
of scarcity originate from an individual's behavior. Also, the mechanisms through which scarcity spurs conict remain debatable. Most
studies argue that scarcity affects productivity and income as well as
income inequality (see references cited in Hsiang et al., 2013). Herein,
we experimentally investigate whether the long-term exposure to resource scarcity increases pastoralists' readiness to engage in antisocial
money-burning behavior, which can be interpreted as a tendency
Corresponding author.
E-mail addresses: sebastian.prediger@giga-hamburg.de (S. Prediger),
bjoern.vollan@uibk.ac.at (B. Vollan), benedikt.herrmann@gmail.com (B. Herrmann).

http://dx.doi.org/10.1016/j.jpubeco.2014.07.007
0047-2727/ 2014 Elsevier B.V. All rights reserved.

towards conict behavior. Scarcity creates competition on the commonly managed grazing lands, and we conjecture that a greater exposure to
scarcity and competition can negatively affect resource users' dispositions towards others and reduce the inhibition threshold to engage in
antisocial acts.
Antisocial or spiteful preferences can be loosely dened as a desire to
harm others at a cost to oneself in the absence of motives of reciprocity.
Related experimental studies show that spiteful behavior is surprisingly
widespread (e.g. Abbink and Sadrieh, 2009; Cason et al., 2002; Fehr
et al., 2008; Herrmann et al., 2008; Herrmann and Orzen, 2008;
Kebede and Zizzo, 2011; Saijo, 1995; Zizzo and Oswald, 2001; Zizzo,
2003), suggesting that an exclusive focus on pro-sociality may come at
the cost of drawing an incomplete picture of human behavior. Reallife examples for antisocial behavior include violence against others,
vandalism and other forms of harmful conict behavior (Abbink and
Sadrieh, 2009). Yet, despite growing evidence for its existence, little is
known about the conditions under which antisocial attitudes are likely
to evolve.
A major challenge to addressing the link between scarcity and antisocial behavior is that an individual's propensity to inict damage on
others is difcult to observe. In natural contexts, often multiple motives
for harmful behavior exist that operate at the same time. This suggests
that an experimental approach is warranted, in which the conict
setting can be controlled and the impact of scarcity can be carefully separated from other potential motives that may evoke antisocial acts. In
order to elicit an individual's readiness to engage in antisocial behavior,

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

we conducted the joy-of-destruction minigame (Abbink and Herrmann,

2011). In this one-shot, two-player game, endowments are equal and
subjects can decide to sacrice income in order to lower another
persons' payoff below one's own. The experimental setup removes all
conventional reasons to engage in such money-burning behavior: No
material gain is achieved by money destruction, no wrongdoing is
punished, no inequality is reduced and anonymity prevents social
comparison and status seeking. Hence, in a narrow interpretation, our
experiment captures antisocial attitudes including a person's readiness
to harm others.
Instead of employing students as a subject pool and inducing scarcity as a function of time, endowment or attempts (see e.g. Shah et al.,
2012), we relate real-life variations in resource scarcity to the experimental play of pastoralists from southern Namibia who subsist on
extensive livestock production on jointly managed grazing lands. We
recruited pastoralists from two geographically proximate areas that
are similar with respect to their political, cultural, religious, climatic
and social background. The main difference between the areas is the
variation in biomass production. This difference is exogenous from
human behavior and stems from geological peculiarities of the study
site. To substantiate and quantify this difference, we use remote sensing
data based on the 23-year average biomass production as a measure for
different degrees of resource scarcity. The design of our study allows us
to examine whether antisocial attitudes are more prevalent among
pastoralists who have been competing for fewer resources for a long
period of time.
We hypothesize a positive relationship between the exposure to
persistent resource scarcity and the incidence of antisocial behavior. In
our study site, the most important channel through which scarcity
may evoke antisocial attitudes is competitive pressure among direct
neighbors: More extreme scarcity on the commons leads to an intensication of competition (Brander and Taylor, 1998; Grossman and
Mendoza, 2003). We believe that the close link between scarcity and
competition is highly relevant in the context of our study site, where
pastoralists share rangelands with fellow villagers, and hence compete
for the same resources within a limited radius. Shleifer (2004) argues
that ethical behavior is often costly and competition may undermine
ethical behavior as it can drive down income and reduce the willingness
to pay for ethical behavior.1 Brandts et al. (2009) recently demonstrate
that the (short-term) exposure to (experimentally induced) competition can negatively affect individuals' disposition towards others. This
in turn may lower the inhibition threshold to engage in harmful
money-burning behavior. Models of social evolutionary theory further
show that spite can be an evolutionary stable trait and postulate a
positive relationship between resource scarcity and the occurrence of
spite (e.g. Gardner and West, 2004; Lehmann et al., 2006, 2009). Scarcity,
the argument goes, creates conditions in which actors may benet from
engaging in costly behaviors that reduce the vital rates of competitors by
harming them, even when the actor faces a cost. This follows from the
fact that spiteful acts can decrease competition and increase the relative
(inclusive) tness (Lehmann et al., 2009). Accordingly, we assume that
people facing a higher degree of real-life scarcity, and hence more intense
competition, are more concerned about outcompeting others and less
reluctant to engage in spiteful acts.
Recently, a number of researchers have begun to empirically study
parochial altruism, i.e. kindness towards members of the own group
and aggressive spite towards members of an out group (e.g. Abbink
et al., 2012; Choi and Bowles, 2007; De Dreu et al., 2012). This literature
highlights, among other things, that exposure to intergroup competition (and hence the existence of a joint opponent) can foster social
cohesion and within-group cooperation. However, it is important to
note that our paper distinguishes from these studies as we are focusing
1
Indeed, Balafoutas et al. (2012) show that sabotage among sportsmen increases especially from less qualied sportsmen when violations of ethical behavior becomes
costless.

on the occurrence of spiteful behavior among people of the same group

(i.e. intragroup spite), and, most importantly, in the absence of preceding intergroup competition and an external opponent whom to blame
for the scarce conditions.2
In line with our ex-ante hypothesis, we nd evidence for a strong
positive relationship between scarcity of natural resources and spite.
One-third of all subjects are willing to reduce a fellow resource users' income at their own cost. This kind of spiteful money-burning behavior
occurs twice as often in areas with greater scarcity, suggesting that a
higher exposure to persistent scarcity and competitive pressure can
evoke antisocial attitudes.
Several aspects arise in interpreting this result as causal. For example, the difference between areas might be due to some unobserved heterogeneity other than resource scarcity, resulting in omitted variable
bias. Further, it could be that people displaying a higher propensity to
engage in conict exploit their resources more aggressively, leading to
degradation (i.e. reverse causality). We address these concerns in
several steps. First, given that variations in resource availability are
due to different soil types in the study region, we argue that reverse
causality is unlikely. The same applies for potential selection bias
through migration within and between areas. Access to grazing lands
is generally restricted and largely depends on the afrmation of traditional authorities and local water user associations. These institutional
arrangements successfully prevent outsiders from intruding the communal lands. In line with that, none of the settlements considered in
this study report migratory movements of livestock or people over the
last decade. Households in the study site are entitled through customary
rights to use the grazing lands around a permanent livestock post that
remains the same over time, but which can get lost if the household
moves elsewhere. The threat of losing access to the commons in combination with poor labor market conditions in the study region provide
strong incentives to stay and continue farming in the assigned area.
Third, all participants share the same ethnicity, are Christians, speak
the same language and live in the same constituency. Hence, variation
in behavior between groups is unlikely to be rooted in cultural, religious
or societal differences, which are frequently stressed as important contextual factors shaping behavior. Finally, our results remain remarkably
robust after employing various controls, including individual beliefs,
absolute income, relative income, within-group inequality, social relationships and share of lifetime spent in the place of residence.
Our paper contributes to three strands of literature. First, there is an
emerging literature that examines the relationship between broader
contextual factors and the evolution of preferences. Experimental
studies in this vein have investigated how inter alia market integration (Henrich et al., 2001), religion (e.g. Henrich et al., 2010), exposure
to different political systems (Brosig-Koch et al., 2011; Ockenfels and
Weimann, 1999), and production technologies (Leibbrandt et al.,
2013) affect behavior and report remarkable evidence in favor of endogenous preference formation. While we are interested in the effects of
long-term scarcity on behavior (i.e. the emergence of conict), others
have begun to investigate the effects of conict on social preferences
(i.e. the consequences of conict). Voors et al. (2012) report that exposure to civil war in Burundi increased altruism, and Bauer et al. (2014)
observe an increase in people's egalitarian motivations towards their
2
Since we focus on persistent long-term exposure to scarcity rather than sudden
changes in the availability of resources, our study also distinguishes from an emerging
body of literature that examines the impact of natural disasters on preferences and behavior (e.g. Cassar et al., 2011; Castillo and Carter, 2011; Chong et al., 2011; Miguel, 2005).
Studies in thin vein have obtained mixed results. For example, Chong et al. (2011) investigate the consequences of the 2010 earthquake in Chile on trust and trustworthiness.
They nd no differences in trust between people who were seriously affected by the earthquake and those who were not affected, but report lower levels of trustworthiness for the
rst mentioned. By contrast, Cassar et al. (2011) nd that Thai victims of the 2004 Asian
Tsunami are substantially more trusting and more risk-averse, as well as more trustworthy than non-victims. A more extreme example for the adverse consequences of natural
disasters are reported by Miguel (2005), who nds a sharp increase in the murder of elderly women in Tanzania after the occurrence of oods or droughts.

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

in-group in the aftermath of war in Georgia and Sierra Leone.3 The novelty of our study is the emphasis on long-term resource scarcity as a potentially important context factor shaping the formation of individual
attitudes. Specically, our results suggest that exposure to persistent
scarcity and competition (in the absence of intergroup conict) can
evoke individuals' propensity to engage in antisocial acts. Second, one
can interpret such spiteful money-burning as a potential tendency towards conict behavior. In this regard, our results are also informative
to research on the onset of conicts and violence.4 A number of macrostudies report a positive relationship between climate induced shocks
and the incidence of violent inter- and intragroup conicts (Burke et al.,
2009; Homer-Dixon, 1999; Miguel et al., 2004; Zhang et al., 2007), although there remain some doubts about the channels through which
this happens (Hsiang et al., 2013). We identify long-term exposure to
greater scarcity and competitive pressure as a source of individual conict
behavior. Lastly, our research is also relevant for the literature on
common-pool resource management. One strand of literature in this
eld has studied the link between resource scarcity and cooperativeness,
with mixed evidence. While some authors posit that collective action is
more likely to evolve under scarcity (see e.g. Ostrom et al., 1999), others
argue that scarcity may drive appropriate competition among users, leading to an ever fast rate of exhaustion (e.g. Grossman and Mendoza, 2003;
Varghese et al., 2013). We contribute to this literature by showing that
antisocial attitudes also need to be taken into account when exploring
the behavioral consequences of scarcity in commons management.
While scarcity may not necessarily undermine subjects' readiness to cooperate, it can negatively affect resource users' dispositions towards
others and evoke tendencies towards conict behavior.
The paper proceeds as follows: The next section provides background information about the study site. Section 3 describes the experimental setup and procedures. The results are presented in Section 4,
followed by a short conclusion in Section 5.
2. Description of the study site
Our study was carried out in the communal lands of the Berseba constituency in the Karas region of southern Namibia. This area is situated
within a semiarid biome and is almost exclusively populated by the
Nama people, an indigenous ethnic group that subsists on extensive
livestock production on commonly managed rangelands. Rainfall is
highly variable and erratic in that area, ranging from 50 to 290 mm
per annum (Kuiper and Meadows, 2002).
2.1. Measure for resource scarcity
We use geographically determined variations in biomass production
as a measure for cross-area differences in resource availability.
Figure A.1 in the supplementary appendix shows the estimated average
seasonal biomass production from 1985 to 2007 for the Karas region,
measured in kilograms per hectare.5 The biomass production varies spatially, ranging from less than 100 kg/ha up to 1200 kg/ha. The study region is split into two areas according to their biomass production: (1) a
high-yield area of 400 kg/ha up to 800 kg/ha and (2) a low-yield area of
below 100 kg/ha.6 According to own calculations based on ofcial
3
By contrast, Cassar et al. (2013) nd that exposure to civil war reinforces kinshipbased norms of morality among Tajiks, while it undermines trust within communities
and impersonal exchange.
4
Although conict behavior exhibited in our experimental setup is comparably harmless.
5
The map was produced by Celeste Espach of the Ministry of Agriculture, Water and
Forestry in Windhoek. The seasonal biomass production was calculated from NOAA/
AVHRR and SPOT/Vegetation satellite images, with the Satellite Monitoring of Arid Rangeland (SMAR) software based on the Monteith model.
6
In cases of prolonged droughts, biomass production can deviate from the average production reported in Figure A.1. Nevertheless, biomass production was always substantially
higher in the high-yield area than in the low-yield area during the last 23 years. Moreover,
in all but two seasons of serious drought (98/99 and 02/03), biomass production was well
above 400 kg/ha in the high-yield area and below 100 kg/ha in the low-yield area.

census data conducted by the Namibian Planning Commission in 2001

(Central Bureau of Statistics, 2004), the population densities are 0.204
and 0.213 persons per km2 in the high-yield area and the low-yield
area, respectively.7 In other words, per-capita resource availability is
higher in the high-yield area. Note that participants' subjective perceptions about resource availability in their areas are in line with
the objective gures. As illustrated in Fig. 1, 63% of the participants
from the high-yield area rate pasture quality and availability as
either good or very good. This stands in stark contrast to the perceptions of people from the low-yield area, where only about 17%
rate pasture quality as good or very good, while more than half
of the sample was of the opinion that their pastures are of poor
or very poor quality (MannWhitney U test: Z = 5.59, p b .001,
n = 119). Moreover, 46% of residents from the high-yield area believe
that the quality of their pastures is better than in most others places in
the Berseba constituency, while only 21% of residents from the lowyield area were of this opinion 52% said that the pasture quality is
worse than in most other places (Z = 2.657, p b .01, n = 100). To
sum up, there is strong evidence that resources are scarcer in the area
we labeled as low yield.
2.2. Exogeneity of resource scarcity in study area
The underlying assumption of our analysis is that the exposure to resource scarcity is largely exogenous to individual behavior. Two important aspects support this claim: First, the areas differ with respect to soil
types and thus geological peculiarities. The high-yield area is part of the
Swartrand terrain, which is characterized by rocky grounds that ease
water inltration and plant growth. In contrast, the low-yield area east
of Berseba contains a different soil type (Witrand) that is seen to be less
favorable for plant growth.8 Therefore, resource differences are not
human made. Second, the possibility for households to move with their
animals between and within the areas is restricted and largely depends
on the afrmation of traditional authorities and (more recently) local
water point associations (WPAs). The traditional authority has the customary right to allocate grazing lands (Republic of Namibia, 2000),
while WPAs which consist of local resource users have the formal
right to grant or deny access to their water resources (Republic of
Namibia, 2004). Participants in our study conrm that migrants have to
ask permission before they can settle and use the grazing lands in a certain area. According to group-interviews held at the end of the experimental sessions, none of the places considered in this study experienced
migratory movements over the last decade. Households typically stay
and keep their animals in permanent livestock posts (plaas) or smaller
settlements (Klocke-Daffa, 2001). A household's entitlement to use the
grazing land around a livestock post remains the same over time, but
will be lost if the household moves elsewhere.9 This also explains why
even in case of drought, families will continue to use their assigned

7
For the calculations, we used the enumeration areas in which the locations we considered in our study were situated and divided the enumeration areas' population by area
size. Visited locations in the high-yield area (low-yield area) are scattered across an area
of 4235 (3800) km2 supporting a population of 868 (811) people. We thank Thimo
Hangula from the National Planning Commission for providing us with the data. For consistency checks, we consulted the local extension ofcers Anton Losper (Ministry of Agriculture, Water and Forestry (MAWF)), Karl Eiseb and Sixtus Isaacks (both Rural Water
Supply (RWS) Keetmanshoop) who operate in the study region. All of them guessed the
population density to be lower in the high-yield area.
8
Pers. com. with Anton Losper, extension ofcer from the MAWF (5th October 2009).
This is also in line with the perception of our participants, of whom 69% were of the opinion that quality is better in the Swartrand terrain than in the Witrand area; only 9% were of
the opposite opinion. The remainder thought there was no difference between the areas.
9
Of course, it happens that single members of a household temporarily move to bigger
settlements or towns, typically for schooling, or in order to look for a wage job. In the
econometric analysis in section 4, we hence control for the length of time individuals have
been staying in their place of residence.

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

Low-yield area

High-yield area

very good good moderate poor very poor

very good good moderate poor very poor

.1

.2

Fraction

.3

.4

.5

Fig. 1. Individual perceptions of pasture quality. Notes: The left-hand and right-hand side of Fig. 1 illustrates the distribution of individual perceptions about pasture quality for low-yield
area and high-yield area residents, respectively. N = 119.
Source: Authors' calculations.

grazing land and eventually sell some of their animals.10 Thus, neither can
newcomers start farming in the communal areas, nor can households
entitled to use a grazing land freely move their livestock to another location. Due to these features, we do not believe that selection bias through
intra- or inter-area migration is a major issue in our study.

2.3. Socioeconomic background

The aim of this paper is to investigate the relationship between scarcity and individuals' propensity to engage in antisocial behavior. In
order to attribute potential differences in behavior between residents
in the low-yield area and high-yield area to the variation in resource
availability, we had to ensure that other important contextual factors
such as cultural, religious, political, climatic or social backgrounds
are very similar across both areas. The study region fullls this precondition: All participants are ethnic Nama, speak Nama, reside in the same
constituency, follow Christianity, and use the same production technology. Inter- and intragroup conicts have not taken place in the study
region since the Nama uprising against German colonizers in 1904.
Moreover, people in both areas are similarly dependent on the availability of natural resources. Wage employment is rare in the whole Berseba
constituency, and extensive livestock production on commonly managed pastures constitutes a key source of income (Central Bureau of
Statistics, 2004). Accordingly, about 81% of the residents in our sample
(n = 120) cite livestock production as their most important source of
income, and 95% possess livestock with an average herd size of 119
small stock units (SSU).11 Probably as a consequence of better resource
availability, people in the high-yield area possess larger herds (mean =
135 SSU, median = 85 SSU) than those in the low-yield area (mean =
10
A questionnaire issued in the low-yield area in 2006 exemplies this. People were
asked Remember the last time when most of the fodder in your grazing area was already
consumed. What did you do? (multiple answers possible). The rst choice for 11 out of 13
farmers was to continue in their assigned grazing area. Nobody stated to move with the
animals elsewhere.
11
Examples for small stock include goats and sheep. Large stock such as cattle, donkeys
and horses is converted into small stock units at the common conversion rate of 1:6.

103 SSU, median = 60 SSU). Since average biomass production is about

four times higher in the high-yield area, we infer that competition for
pasture resources is more intense among pastoralists residing in the
low-yield area. Differences in grazing availability seem to also affect
herd composition. Livestock owners in the high-yield area keep signicantly more cattle and sheep (see Table A.1 in the Appendix). Both
species are more demanding with respect to fodder requirements
than goats. People typically send their livestock in a radius of about
6 km around their houses and water points. If not used for own consumption, animals are mainly sold on auctions that are held in regular
intervals in Berseba and other bigger settlements scattered across the
area. Livestock is not only an important source of income and food,
but it is also indicative of wealth, fullls an insurance function and represents social status in the Nama culture (Klocke-Daffa, 2001). In the regression analyses reported in Section 4, we use individual livestock
numbers as a proxy for income in order to assess the impact of poverty
or inequality on the occurrence of spite.
Sixty-ve percent of the participants were male and had attended
school for about 7.9 years. Participants were 42 years old on average
and have been continuously living in their livestock posts (or settlements) for an average of 22 years. These and further sample characteristics are presented in Table A.1 in the Appendix.

3. Experimental design and procedures

We carried out 20 experimental sessions in 15 locations, with 10 sessions being held in each (high-yield and low-yield) area. Each session
consisted of 6 participants, resulting in a total sample size of 120. A session consisted of four tasks: (1) a joy-of-destruction minigame, (2) a
public goods experiment (not reported here),12 (3) an individual
12
To account for order effects, we alternated the sequence in which the experiments
were held half of all sessions (5 in each area) started with the JoD experiment, half with
the public goods (PG) experiment. We control for sequencing effects and individual
contribution decisions in the PG experiment in the regression analyses in section 4 (see
Table 2). A detailed analysis of the results of the PG experiment is provided in Prediger
et al. (2013).

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

follow-up questionnaire and (4) a short group discussion on migration,

internal conicts and environmental problems at the very end of the
session. All sessions were conducted in the dry season in November
December 2009.
3.1. The joy-of-destruction experiment
The design of the joy-of-destruction (JoD) minigame (Abbink and
Herrmann, 2011) offers a simple way to analyze antisocial or conict
behavior, which we dene as a willingness to harm others in the absence of motives of reciprocity even if this comes at a cost to oneself.
In the experiment, two subjects were randomly matched. Both received
an initial endowment of N$10 (equivalent to PPP US$1.6) and had to
decide whether or not to reduce the other player's income by N$5 at a
personal cost of N$1.13 Thus, destruction entails no material benets
for the destroying subject but a personal sacrice. Unlike in money
burning experiments (e.g. Zizzo and Oswald, 2001), inequity aversion
or envy are removed as potential motives for money burning. Because
it is played one-shot and decisions are made anonymously, strategic aspects should not matter either (Cubitt et al., 2011) and not burning is
the strictly dominant strategy of a rational self-concerned player. The experiment has three possible outcomes, which are summarized in Table 1.
First, both subjects decide not to reduce the other's income. In that case,
each subject remains with N$10. Second, each subject decides to destroy
the other's income, leaving both with N$4. Third, one subject reduces the
other's return while the other does not, resulting in an unequal payoff distribution of N$9 for the destroying party and N$5 for the victim
of destruction. After a subject had made a decision, the subject was
asked to state his or her belief about the interaction partner's choice.
To avoid potential changes in the relationship between behavior and
beliefs (Gchter and Renner, 2010), we refrained from incentivizing
beliefs that is, subjects were not rewarded for guessing correctly.
3.2. Experimental procedures
The standard procedures of economic experiments were applied:
Communication was strictly prohibited, the decisions were made anonymously and the identity of the interaction partner(s) was never
disclosed to the participants. We used uniform instructions that were
translated from English into Afrikaans (and back to English) and
presented orally by a local eld assistant. We employed the same eld
assistant for all 20 sessions. Subjects conducted the experiments with
pen and paper and were paid individually and privately at the end of
the entire session for both games. We did not provide subjects any
feed-back between the two games. They earned on average N$18.1
( PPP US$3), including a show-up fee of N$10 (but excluding the
money they earned in the public goods experiment). In comparison,
the daily salary of a wageworker in the study region amounts to about
N$40.
3.3. Recruitment
Most locations in the study region are livestock posts that are
far-scattered within the territory. Livestock posts typically comprise
four to eight houses. To avoid having more than one participant from
the same household, we only considered livestock posts with six to
eight houses. Thirteen locations fullled this requirement and were
used in this study. The remaining seven sessions were held in two
rural settlements: four sessions in Kutenhoas (in the high-yield area
and comprised of 32 households) and three in Snyfontain (in the low-

13
Subjects were asked Do you want to pay one dollar to reduce your partner's income
by 5 dollars? In the following, however, we use the terms burn, destroy or reduce
interchangeably.

Table 1
Payoff table for the joy-of-destruction experiment.
Source: Authors' calculation.

Player A
Player B

Reduce

Not reduce

Reduce
Not reduce

4/4
5/9

9/5
10/10

yield area and comprised of 27 households).14 To minimize the confounding effects of cross talk in these two settlements, all sessions
took place the same day and participants were recruited immediately
before the start of the session.15 We invited one person per household
to participate. Where there were more than 6 households per location,
we randomly determined which households could send a member for
participation.
4. Results
Thirty-two percent of subjects decided to reduce their partner's
income, although this came at a cost to them. This clearly contradicts
the predictions of the homo economicus model, but it is in line with
results from related studies (e.g. Abbink and Sadrieh, 2009; Abbink
and Herrmann, 2011). In a repeatedly played joy-of-destruction game,
for example, Abbink and Sadrieh (2009) report destruction rates up to
40% among Dutch students. Money burning happened more frequently
in the low-yield area. There, 40% of all subjects diminished their
partner's income, compared to 23.3% in the high-yield area (Fisher's
exact test, p = 0.04, one-sided, n = 120).
Table 2 displays the results of linear probability regressions where
we regress subject's burning decision (y = 1 if subject burns money)
on area of residence, beliefs, absolute and relative income and further
covariates.16 In line with our hypothesis and descriptive results, we
nd a statistically signicant and economically sizeable difference between areas in the incidence of conict, i.e. antisocial money-burning
behavior. In the rst model of Table 2, the probability for destroying
money increases by 17 percentage points if the subject was from the
area of greater scarcity. The area effect remains signicant across all
specications.
Result 1: Harmful money-burning occurs signicantly more frequently in the low-yield area.
Unsurprisingly, we further nd individuals' beliefs about the other
player's behavior to have very strong predictive power for burning decisions: The probability of choosing to reduce another person's income increases by 66 percentage points if the subject expected him- or herself
to suffer a destruction of income. Most importantly, however, controlling for beliefs does not remove the signicant area difference on the
contrary, the area effect becomes bigger and more signicant. Twentysix percent of the sample had negative beliefs, and among them, 81%
reduced the other's income. This closely resembles the results of
Abbink and Herrmann (2011), who report that 86% of the money
burners had negative beliefs. In our sample, this kind of preemptive
retaliation occurs more frequently in the low-yield area. Of those
expecting to become victims of destruction, 93% destroyed money in

14
The frequency of burning decisions does not differ between livestock posts and settlements (Chi2 = 0.05, p = 0.9, n = 120), and hence does not seem to be a question of village
size.
15
Results from Chi-square tests provide no evidence of cross-talk or contagion effects in
Snyfontain (Chi2: 1.3, p = 0.51, n = 18) or Kutenhoas (Chi2 = 2.4, p = 0.49, n = 24).
16
We qualitatively obtain very similar results if we use probit models (see Table A.2 in
the Appendix). Note further that the statistical signicance of the area effect is larger in
the probit specications. This also applies to all robustness tests reported in the supplementary appendix.

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

Table 2
Determinants of individual money burning decisions.
Source: Authors' calculation.

(1)
Low-yield area

0.167**
(0.074)

(2)
0.178***
(0.057)

(3)
0.153**
(0.071)

(4)
0.193**
(0.089)

(5)
0.065
(0.095)

(6)
0.138*
(0.076)

(7)
0.145*
(0.078)

Perception of poor pasture quality

Negative belief

0.664***
(0.085)

Socio-demographics
Male
Age
Education
Farmer
Length of stay

0.675***
(0.087)

0.665***
(0.094)

0.012
(0.077)
0.000
(0.003)
0.006
(0.019)
0.063
(0.074)
0.053
(0.095)

0.020
(0.079)
0.001
(0.003)
0.007
(0.019)
0.059
(0.076)
0.056
(0.096)

0.030
(0.074)
0.001
(0.003)
0.011
(0.018)
0.085
(0.075)
0.277*
(0.148)
0.405**
(0.174)

0.0004*
(0.0002)

0.0003
(0.0002)
0.0004
(0.0004)

0.0003*
(0.0002)

Length of stay low Area

Economic situation
Abs. Herd size

0.668***
(0.089)

Abs. Herd size low Area

0.674***
(0.084)
0.009
(0.078)
0.001
(0.003)
0.004
(0.019)
0.074
(0.075)
0.051
(0.095)

Number of family members

Conict with others
Other controls
JoD rst
Contribution in PG
Constant
Observations
R2
F
p
Adjusted R2

0.233***
(0.043)
120
0.032
5.011
0.037
0.0239

0.056
(0.037)
120
0.423
93.97
0
0.413

0.016
(0.078)
0.001
(0.003)
0.005
(0.019)
0.078
(0.075)
0.064
(0.097)

0.010
(0.075)
0.000
(0.003)
0.003
(0.019)
0.034
(0.068)
0.047
(0.094)

0.0001
(0.0002)

0.0001
(0.0002)

0.0003
(0.0002)

0.0003
(0.0002)

0.021*
(0.012)
0.043
(0.038)
0.0003
(0.0002)

0.025
(0.018)
0.019
(0.022)
0.064
(0.092)

0.026
(0.019)
0.020
(0.022)
0.050
(0.083)

0.029
(0.019)
0.020
(0.020)
0.042
(0.095)

0.024
(0.018)
0.021
(0.022)
0.088
(0.093)

0.027
(0.019)
0.021
(0.022)
0.099
(0.094)

0.019
(0.019)
0.017
(0.021)
0.047
(0.099)

0.040
(0.056)
0.088
(0.109)
0.333
(0.268)
120
0.476
86.49
0
0.407

0.043
(0.054)
0.090
(0.111)
0.330
(0.271)
120
0.479
147.4
0
0.404

0.040
(0.055)
0.068
(0.104)
0.448
(0.276)
120
0.497
132.2
0
0.424

0.036
(0.060)
0.067
(0.106)
0.346
(0.267)
120
0.478
106.2
0
0.408

0.032
(0.060)
0.075
(0.108)
0.366
(0.280)
120
0.482
119.8
0
0.408

0.060
(0.070)
0.079
(0.109)
0.341
(0.289)
119
0.490
44.54
0
0.421

Rel. Herd size low Area

Social relationships
Number of friends

0.672***
(0.085)

0.206**
(0.080)
0.651***
(0.084)

0.0003
(0.0002)

0.027**
(0.010)

Rel. Herd size

Inequality in herd size

(8)

0.0002
(0.0002)

Notes: Linear probability models. Dependent variable: Burning decision in the JoD experiment (burn = 1, not burn = 0). The number of observations included in regression 8 deviates
from the actual sample size due to 1 missing values in Perception of poor pasture quality. Regression estimations are reported with heteroskedasticity-robust standard errors clustered
at session level (in parentheses). ***, **, and * indicates statistical signicance at the 1%, 5% and 10% level, respectively.

the low-yield area, compared to 69% in the high-yield area. When analyzing the behavior of subjects who did not expect their counterparts to
reduce their income, we nd that in the resource-scarce area 22%
burned money nevertheless, compared to 7% in the high-yield area.
We interpret such behavior as a clear indication of negative dispositions
towards fellow resource users. If including an interaction term between
negative beliefs and the area dummy in the regression analysis, we observe that this difference is signicant (t = 1.93, p = 0.06) between
areas (regression output not shown), conrming that negative dispositions are more prevalent among low-yield area residents.
We also account for the possibility that social relationships to other
group members may affect decision making. The negative signs of
number of friends and number of extended family member in estimations
3 through 8 suggest that the more friends and family members in the
same group, the lower the probability of destroying another group
member's money although the effects are not signicant at

conventional levels. The same applies to conicts with other group

members17 and observable socio-demographic characteristics. From
Table 2, it further becomes apparent that subjects whose main source
of income is farming do not behave differently from wageworkers and
pensioners, who constitute the reference category. We employ the variables Length of stay as well as an interaction term between this variable
and the area dummy (estimation 5) as further controls. The variable is
the quotient of the number of years a participant has been continuously
living in her place of residence and her age. Hence, the variable ranges
from 0 (if participant just moved back to his or her place of residence)
to 1 (if participant had spent his/her entire life in the place of residence).

17
We also gathered information about disputes at village level by asking participants
during a postgame group discussion whether there had been any unsolved conicts
among residents. This was abnegated in all sessions.

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

We observe no difference between areas in the behavior of people who

only recently moved to their family's livestock post (see low-yield area
in estimation 5). However, we nd evidence for heterogeneous effects
across areas for those subjects who have been staying in their village
for a longer period of time. Residents of the area with greater scarcity
tend to be more likely to burn money the longer they have been residing
in their place (as indicated by the interaction term). The opposite
relationship is found for high-yield area residents. This observation is
consistent with the idea that attitudes towards others evolve over
time and further suggests that positive attitudes suffer from exposure
to persistent scarcity and competition.
As mentioned in Section 3, we also conducted a one-shot linear public goods (PG) experiment with the participants of our study. Although
this game does not occupy center stage herein, it is important in exploring whether sequencing effects and contribution decisions made in the
PG game affect money-burning behavior. JoD rst is a binary variable
that takes the value of 1 if the JoD experiment was conducted before
the PG experiment and 0 otherwise. From estimations 3 through 8 it
becomes visible that the chronological order in which the experiments
were held had no effect on the incidence of spite. Interestingly, contribution decisions in the PG game (Contribution in PG) are not correlated
with experimental play in the JoD experiment, either. This is because
about 16% of the subjects in our sample burned money in the JoD
experiment but made above median contribution in the PG experiment
(n = 19). This observation is interesting and a bit surprising at once.
However, there are other authors reporting a coexistence of antisocial
and prosocial behavior within individuals across different decision situations, too (e.g. Abbink et al., 2010; Bardsley, 2008; Herrmann and
Orzen, 2008; Sadrieh and Schrder, 2012), which indicates that subjects' behavior can strongly depend on the experimental environments
in which they are acting (Bowles, 2008). The design of our study does
not allow to give an ultimate answer to the question why some people
behave antisocially in the JoD experiment but not necessarily less cooperatively in the PG experiment.18 However, it is important to note that
this observation does not challenge our main result. If we run a robustness test where we exclude all those subjects from the regression who
burned money in the JoD experiment and made above median contributions in the PG experiment (n = 19), we still obtain a highly signicant
difference in the incidence of spite across areas (see Table A.3 in the
Appendix).
Greater scarcity could result in higher inequality and poverty, which
both may affect players' propensity to engage in antisocial behavior.19
We hence explore whether the striking difference observed across
areas is mainly driven by differences in inequality or absolute and relative income poverty rather than scarcity. As a proxy for absolute income,
we use subjects' total herd size, measured in terms of SSU (small stock
units). Regression 3 shows that people with larger herds tend to be
less likely to destroy money, but the size effect is tiny. In order to investigate the impact of a subject's relative income position within his or her
group, we calculated the relative distance of each player's income (again
in terms of herd size) from the average income of the other session
members.20 As expected, the relatively better-off are less likely to
burn money than fellow villagers who are at the bottom of the income
18
The literature considers framing effects and experimenter demand effects as potential
explanations for changes in individual behavior across different experimental setups (see
e.g. Bardsley, 2008; Cubitt et al., 2011; Dreber et al., 2013; Dufwenberg et al., 2011; Tversky
and Kahneman, 1981; Zizzo, 2010). While demand effects might affect the absolute frequency of money burning behavior in the JoD experiment, they are unlikely to drive differences in behavior between areas on which we focus herein. A more detailed discussion
about the relevance of farming effects in our study can be found in Appendix B.
19
There is some debate about whether or not one should control for income as this
might bias the coefcient of interest (Hsiang et al., 2013). Hence, we present our regression with and without controlling for income effects.
20
We obtain qualitatively the same results if we divide groups according to herd size into terciles and use these variables as proxies for a subject's relative income position within
his or her relevant comparison group. People in the upper tercile are signicantly less likely to engage in money burning behavior (results available upon request).

distribution (estimation 6). From this observation, we infer that positional concerns mattered indeed for the choice of money burning. Yet,
the coefcient of the area dummy remains signicant and decreases
from 17.7% (estimation 2) to 13.8% (estimation 6). Thus, only about
20% of the scarcity effect seems to be due to income and income inequality. Note further that we nd no evidence for interaction effects
between absolute herd size and low-yield area (estimation 4) and relative
herd size and the area dummy (estimation 7), suggesting that absolute
and relative income poverty have similar effects in both areas. Finally,
we account for the possibility that within group inequality (measured
by the standard deviation of herd size at group level) could affect
behavior. However, counter to our prior expectations, we do not observe a signicant correlation with money-burning behavior.
Apparently, herd size is a fairly good proxy for farmers' income, but
not for that of wageworkers and pensioners. For the latter two groups,
livestock production typically constitutes a secondary source of income.
As a robustness test, we thus run the same regressions as in Table 2 for a
sample restricted to farmers (n = 97, Table A.4 in the Appendix). When
doing so, it turns out that neither relative nor absolute herd size significantly impacts the probability of money burning, while farmers in more
unequal groups tend to be less likely to destroy their interaction
partner's experimental income. Since the area coefcient remains highly signicant if employing these controls, we infer that income poverty
or inequality alone do not explain the higher occurrence of moneyburning behavior among residents in the low-yield area.
Result 2: The higher incidence of antisocial behavior observed in the
area with greater scarcity is not caused by cross-area differences in inequality, absolute or relative poverty.
Finally, in estimation 8 of Table 2, we replace low-yield area with a
variable that captures individuals' perception about the quality of their
pasture. This variable is highly correlated with the area dummy ( =
.44, p b .001). It takes the value of 1 if the subject was of the opinion
that the pasture is of poor or very poor quality and 0 otherwise.
We observe a higher incidence of antisocial behavior (i.e., burning decisions) among those who had negative perceptions about the state of
their resource base. Separate regressions for each area reveal that this
tends to be the case for both areas (see Table A.5 in the Appendix). In
the high-yield area, only 6 out of 59 subjects were of the opinion that
their pastures are of poor or very poor quality. Among those 6 subjects, 3 decided to reduce their partner's income. In the low-yield area,
the majority (51.7%) assessed the quality as poor or very poor and
45% of them were willing to destroy money. These ndings provide suggestive evidence in favor of our claim that it is indeed the exposure to
resource scarcity that increases subjects' readiness to engage in conict
behavior, i.e. harmful money-burning behavior.
5. Concluding remarks
The world's natural resources are increasingly in short supply.
Climate change and population growth, among others, are likely to
exacerbate the problem of increasing scarcity in future. Given that the
ongoing depletion of natural resources constitutes a serious threat to
the livelihoods of millions of people especially in developing countries
(World Bank, 2009), a natural question is how scarcity shapes attitudes
and behavior of affected individuals. Herein, we tried to tackle this question by experimentally examining whether differences in long-term
exposure to resource scarcity on the commons affect Namibian pastoralists' readiness to engage in spiteful money-burning behavior. As a
measure for the degree of scarcity, we used exogenous variations in
average biomass production in the research region. These cross-area
differences in the availability of grazing resources stem from geological
peculiarities of the study site, which makes reverse causality highly
unlikely.
We report initial evidence for a strongly positive relationship between resource scarcity and spite in the absence of intergroup conict.
In the area where scarcity and competitive pressure are more prevalent,

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

almost twice as many subjects displayed a readiness to destroy their fellow resource users' income at a cost to themselves. We interpret this
form of harmful money-burning as an individual's tendency towards
conict behavior. Participants in our study share a similar political,
cultural, religious and societal background. Hence, it is very unlikely
that cross-area differences in behavior are driven by these or other
omitted factors. Surprisingly, the higher incidence of spite in the
resource-scarce area does not seem to be rooted in income poverty,
income inequality or other observable socioeconomic characteristics,
either. Although better-off participants tend to be less likely to diminish
another person's income, the area effect remains large and signicant
after employing controls for absolute and relative income in the econometric analysis. We further nd that subjects' perceptions about pasture
quality are highly and positively correlated with money burning decisions. Specically, those who perceive their pasture's quality as poor
were signicantly more likely to burn money than those who considered the quality to be good. We cannot entirely rule out that the
strong area effect we observe is due to some unobservable heterogeneity in other than exposure to scarcity and competition. However, the
evidence presented herein strongly suggests that it is indeed the greater
degree of scarcity and competitive pressure that makes people in the
low-yield area more likely to behave spitefully towards fellow resource
users.
Our results are in line with models in psychology (e.g. Jackson, 1993)
and social evolutionary biology (e.g. Lehmann et al., 2009). Evolutionary
models argue that harmful spite can be evolutionarily favorable, particularly in periods of scarcity and high competition, as it increases the
agent's tness relative to the harmed competitor. Furthermore, psychological research shows that persons selectively perceive those social
objectives that are most relevant to currently salient roles (McCall and
Simmons, 1978) and that scarcity alters the way in which people
allocate their attention, leading them to engage more deeply in some
problems while neglecting others (Shah et al., 2012). Related to our
research, it seems that people who are exposed to a higher degree of
scarcity (i.e. live in a more competitive environment), and who therefore need to try harder to sustain their livelihood, are more inclined to
perceive their interaction partner as a competitor against whom they
have to prevail. This perception may result in a neglect of efciency
concerns and a stronger desire to come out ahead or to harm others.
A natural question to ask is how adequately experimental play reects participants' real attitudes. With the data at hand, we are not
able to answer this question. However, there are several studies demonstrating the external validity of experimental behavior (e.g. Benz and
Meier, 2008; Carpenter and Seki, 2011; Fehr and Leibbrandt, 2011;
Rustagi et al., 2010). Rustagi et al. (2010), for example, show that
Ethiopian forest user groups with a larger share of conditional cooperators (as measured in public goods experiments) are more successful in
forest management. Similarly, Fehr and Leibbrandt (2011) nd that
Brazilian shermen who exhibit more cooperative behavior in a laboratory experiment are less likely to exploit common-pool resources. So
far, evidence on the external validity of laboratory preference measures
is restricted to the prosocial domain of human decision making. Investigating whether external validity applies to the antisocial domain as
well, remains an important and interesting avenue for future research.
Our results suggest that long-term exposure to scarcity and competition on the commons can lower resource users' inhibition threshold to
engage in harmful behavior. We observe such tendencies in an experimental setting that allows people to destroy a fellow resource users'
(experimental) income without taking the risk of being detected and
sanctioned. In real-life, however, formal and informal institutions may
prevent such tendencies towards conict behavior. The results of our
study allow for some cautious policy implications. If competition drives
spiteful behavior and if the occurrence of spite is positively correlated
with conict in real life, policies should aim at efcient resource use
and at providing outside options to people, thereby reducing dependence from the limited resources at place. Presumably the most efcient

way to increase the number of outside options to local scarcity of

resource is public investments into education and training. Better educated and trained people have not to rely on use of local land, but can
make use of other, as satisfying options to work elsewhere. Another
policy that could reduce the resource pressure, and which is currently
pursued by the Government in some parts of Namibia, is resettling
farmers within a land reform program. Lastly, if scarcity is inevitable,
conicts might best be mitigated by institutions that facilitate communication and conict resolution among users of natural resources
(Ostrom, 1990).
Acknowledgments
This research is part of the BIOTA Southern Africa Project funded by
the German Federal Ministry of Education and Research (Commission
number: 01 LC 0024A). We thank Richard Isaacks for his great eld
assistance and all participants for their time and hospitality. We are
grateful to Celeste Espach (Namibian Ministry of Agriculture, Forestry
and Water) for providing us with data and maps on seasonal biomass
production in the study region and to Thimo Hangula (Namibian Statistical Agency, Namibia) for providing us with population data for the
study region. Many thanks also go to Leon Lubbe, Anton Losper (both
Ministry of Agriculture, Forestry and Water), Karl Eiseb and Sixtus
Isaacks (both Rural Water Supply Keetmanshoop) for sharing their
profound knowledge about the local socioeconomic and ecological condition with us. The paper greatly beneted from comments on an earlier
version of Loukas Balafoutas, Michael Baurmann, Esther Blanco, Tim
Cason, Carsten De Dreu, Stefanie Engel, Antonio M. Espn Martn,
Simon Gchter, Renate Hartwig, Michael Kirk, Evelyn Korn, Andreas
Landmann, Alexander Libman, David Rand and two anonymous reviewers as well as participants of the PegNet Conference 2012, IMEBE
Conference 2013, and Workshop participants in Exeter, Innsbruck,
Zrich, Cologne, Bochum, Hamburg, Bielefeld and Oslo.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.jpubeco.2014.07.007.
References
Abbink, K., Herrmann, B., 2011. The moral costs of nastiness. Econ. Inq. 49 (2), 631633.
Abbink, K., Sadrieh, A., 2009. The pleasure of being nasty. Econ. Lett. 105 (3), 306308.
Abbink, K., Brandts, J., Herrmann, B., Orzen, H., 2010. Intergroup conict and intra-group
punishment in an experimental contest game. Am. Econ. Rev. 100 (1), 420427.
Abbink, K., Brandts, J., Herrmann, B., Orzen, H., 2012. Parochial altruism in inter-group
conicts. Econ. Lett. 117 (1), 4548.
Balafoutas, L., Lindner, F., Sutter, M., 2012. Sabotage in tournaments: evidence from a
natural experiment. Kyklos 65 (4), 425441.
Bardsley, N., 2008. Dictator game giving: altruism or artefact? Exp. Econ. 11 (2), 122133.
Bauer, M.,Cassar, A.,Chytilov, J.,Henrich, J., 2014. War's enduring effects on the development of egalitarian motivations and in-group biases. Psychol. Sci. 25 (1), 4757.
Benz, M.,Meier, S., 2008. Do people behave in experiments as in the eld? Evidence from
donations. Exp. Econ. 11 (3), 268281.
Bowles, S., 2008. Policies designed for self-interested citizens may undermine the moral
sentiments: evidence from economic experiments. Science 320 (5883), 16051609.
Brander, J., Taylor, S.M., 1998. The simple economics of Easter Island: a RicardoMalthus
model of renewable resource use. Am. Econ. Rev. 88, 119138.
Brandts, J., Riedl, A., van Winden, F., 2009. Competitive rivalry, social disposition, and
subjective well-being: an experiment. J. Public Econ. 93 (1112), 11581167.
Brosig-Koch, J., Helbach, C., Ockenfels, A., Weimann, J., 2011. Still different after all these
years: solidarity behavior in East and West Germany. J. Public Econ. 95, 13731376.
Brunnschweiler, C.N., Bulte, E.H., 2009. Natural resources and violent conict: resource
abundance, dependence, and the onset of civil wars. Oxf. Econ. Pap. 61 (4), 651674.
Burke, M.B.,Miguel, E.,Satyanath, S.,Dykema, A.,Lobell, D.A., 2009. Warming increases the
risk of civil war in Africa. Proc. Natl. Acad. Sci. 106 (49), 2067020674.
Carpenter, J.P., Seki, E., 2011. Do social preferences increase productivity? Field experimental evidence from shermen in Toyama Bay. Econ. Inq. 49 (2), 612630.
Cason, T., Saijo, T., Yamato, T., 2002. Voluntary participation and spite in public good
provision experiments: an international comparison. Exp. Econ. 5 (2), 133153.
Cassar, A., Healy, A., von Kessler, C., 2011. Trust, risk, and time preferences after a natural
disaster: experimental evidence from Thailand. Working PaperUniversity of San
Francisco.

S. Prediger et al. / Journal of Public Economics 119 (2014) 19

Cassar, A.,Grosjean, P.,Whitt, S., 2013. Legacies of violence: trust and market development.
J. Econ. Growth 18 (3), 285318.
Castillo, M., Carter, M., 2011. Behavioral Responses to Natural Disasters. George Mason
University Discussion Paper, Interdisciplinary Center for Economic Science.
Central Bureau of Statistics (CBS), 2004. 2001 Population and Housing Census. National
Planning Commission, Karas Region, Windhoek.
Choi, J.-K., Bowles, S., 2007. The coevolution of parochial altruism and war. Science 308,
636640.
Chong, A., Fleming, D., Hernn, B., 2011. Trust and Trustworthiness in the Aftermath of
Natural Disasters: Experimental Evidence from the 2010 Chilean Earthquake.
Institute for International Economic Policy (Mimeo).
Collier, P., Hoefer, A., 1998. On economic causes of civil war. Oxf. Econ. Pap. 50 (4),
563573.
Cubitt, R.P.,Drouvelis, M., Gchter, S., 2011. Framing and free riding: emotional responses
and punishment in social dilemma games. Exp. Econ. 14 (2), 254272.
De Dreu, C.K.W.,Greer, L.L.,Handgraaf, M.J.J.,Shalvi, S.,Van Kleef, G.A., 2012. Oxytocin modulates selection of allies in intergroup conict. Proc. R. Soc. B Biol. Sci. 279 (1731),
11501154.
Dreber, A., Ellingsen, T., Johannesson, M., Rand, D.G., 2013. Do people care about social
context? Framing effects in dictator games. Exp. Econ. 16 (3), 349371.
Dufwenberg, M., Gchter, S., Hennig-Schmidt, H., 2011. The framing of games and the
psychology of play. Games Econ. Behav. 73, 459478.
Fehr, E., Leibbrandt, A., 2011. A eld study on cooperativeness and impatience in the
tragedy of the commons. J. Public Econ. 95 (910), 11441155.
Fehr, E.,Hoff, K.,Kshetramade, M., 2008. Spite and development. Am. Econ. Rev. Pap. Proc.
98 (2), 494499.
Gchter, S.,Renner, E., 2010. The effects of (incentivized) belief elicitation in public goods
experiments. Exp. Econ. 13, 364377.
Gardner, A., West, S.A., 2004. Spite and the scale of competition. J. Evol. Biol. 17,
11951203.
Grossman, H.I., Mendoza, J., 2003. Scarcity and appropriative competition. Eur. J. Polit.
Econ. 19, 747758.
Henrich, J.,Boyd, R.,Bowles, S.,Camerer, C.,Fehr, E.,Gintis, H.,McElreath, R., 2001. In search
of homo economicus: behavioral experiments in 15 small-scale societies. Am. Econ.
Rev. Pap. Proc. 91 (2), 7378.
Henrich, J., Ensminger, J., McElreath, R., Barr, A., Barrett, C., Bolyanatz, A., Cardenas, J.C.,
Gurven, M., Gwako, E., Henrich, N., Lesorogol, C., Marlowe, F., Tracer, D., Ziker, J.,
2010. Markets, religion, community size, and the evolution of fairness and punishment.
Science 327 (5972), 14801484.
Herrmann, B.,Orzen, H., 2008. The appearance of homo rivalis: social preferences and the
nature of rent seeking. CeDEx Discussion Paper 200810University of Nottingham.
Herrmann, B., Thni, C., Gchter, S., 2008. Antisocial punishment across societies. Science
319 (5868), 13621367.
Homer-Dixon, T.F., 1999. Environment, Scarcity, and Violence. Princeton University Press,
Princeton.
Hsiang, S.M., Burke, M., Miguel, E., 2013. Quantifying the inuence of climate on human
conict. Science 341 (6151). http://dx.doi.org/10.1126/science.1235367.
Jackson, J.W., 1993. Realistic group conict theory: a review and evaluation of the theoretical
and empirical literature. Psychol. Rec. 43 (3), 395415.
Kebede, B.,Zizzo, D.J., 2011. Envy and agricultural innovation: an experimental case study
from Ethiopia. Centre for Behavioural and Experimental Social Science (CBESS)
Discussion Paperpp. 1112.

Klocke-Daffa, S., 2001. Wenn du hast, mut du geben: Soziale Sicherung im Ritus und
Alltag bei den Nama von Berseba/Namibia. LIT, Mnster.
Kuiper, S.M., Meadows, M.E., 2002. Sustainability of livestock farming in the communal
lands of Southern Namibia. Land Degrad. Dev. 13, 115.
Lehmann, L., Bargum, K., Reuter, M., 2006. An evolutionary analysis of the relationship
between spite and altruism. J. Evol. Biol. 19 (5), 15071515.
Lehmann, L.,Feldman, M.W.,Rousset, F., 2009. On the evolution of harming and recognition in
nite panmictic and innite structured populations. Evolution 63 (11), 28962913.
Leibbrandt, A.,Gneezy, U.,List, J.A., 2013. Rise and fall of competitiveness in individualistic
and collectivistic societies. Proc. Natl. Acad. Sci. 101 (23), 93059308.
McCall, G.J., Simmons, J.L., 1978. Identities and Interactions. Free Press, New York.
Miguel, E., 2005. Poverty and witch killing. Rev. Econ. Stud. 72 (4), 11531172.
Miguel, E., Satyanath, S., Sergenti, E., 2004. Economic shocks and civil conict: an instrumental variables approach. J. Polit. Econ. 112 (4), 725753.
Ockenfels, A.,Weimann, J., 1999. Types and patterns: an experimental Eastwest-German
comparison of cooperation and solidarity. J. Public Econ. 71 (2), 275287.
Ostrom, E., 1990. Governing the Commons: The Evolution of Institutions for Collective
Action. Cambridge University Press.
Ostrom, E.,Burger, J.,Field, C.B.,Norgaard, R.B.,Policansky, D., 1999. Revisiting the commons:
local lessons, global challenges. Science 284 (5412), 278282.
Prediger, S., Vollan, B., Herrmann, B., 2013. Resource scarcity, spite and cooperation.
Working Papers in Economics and Statistics 201310University of Innsbruck.
Republic of Namibia, 2000. Traditional Authorities Act, 2000 (Windhoek).
Republic of Namibia, 2004. Water Resources Management Act No. 24 (Windhoek).
Rustagi, D., Engel, S., Kosfeld, M., 2010. Conditional cooperation and costly monitoring
explain success in forest commons management. Science 330 (6006), 961965.
Sadrieh, A., Schrder, M., 2012. The desire to inuence others. FEMM Working Paper No.
12027.
Saijo, T., 1995. The spite dilemma in voluntary contribution mechanism experiments. J.
Con. Archaeol. 39 (3), 535560.
Shah, A.K., Mullainathan, S., Shar, E., 2012. Some consequences of having too little.
Science 338 (6107), 682685.
Shleifer, A., 2004. Does competition destroy ethical behavior? Am. Econ. Rev. Pap. Proc. 94
(2), 414418.
Tversky, A., Kahneman, D., 1981. The framing of decisions and the psychology of choice.
Science 211 (4881), 453458.
Varghese, S.K.,Veettil, P.K.,Speelman, S.,Buysse, J.,Van Huylenbroeck, G., 2013. Estimating
the causal effect of water scarcity on the groundwater use efciency of rice farming in
South India. Ecol. Econ. 86 (1), 5564.
Voors, M.,Nillesen, E.,Verwimp, P.,Bulte, E.,Lensink, R.,van Soest, D., 2012. Violent conict
and behavior: a eld experiment in Burundi. Am. Econ. Rev. 102 (2), 941964.
World Bank, 2009. World Development Report 2010: Climate Change and Development.
The World Bank, Washington D.C.
Zhang, D.D., Zhang, J., Lee, H.F., He, Y., 2007. Climate change and war frequency in eastern
China over the last millennium. Hum. Ecol. 35 (4), 403414.
Zizzo, D.J., 2003. Money burning and rank egalitarianism with random dictators. Econ.
Lett. 81 (2), 263266.
Zizzo, D.J., 2010. Experimenter demand effects in economic experiments. Exp. Econ. 13
(1), 7598.
Zizzo, D.J., Oswald, A., 2001. Are people willing to pay to reduce others' incomes? Ann.
Econ. Stat. 6364, 3965.