See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/276941364
Why do conﬂicts over scarce renewable resources turn violent? A qualitative
comparative analysis
Article  in  Global Environmental Change · May 2015
DOI: 10.1016/j.gloenvcha.2015.04.008
CITATIONS
48
READS
208
1 author:
Some of the authors of this publication are also working on these related projects:
Environmental Peacemaking? View project
The link between climate change and conflict View project
Tobias Ide
University of Melbourne
49 PUBLICATIONS   480 CITATIONS   
SEE PROFILE
All content following this page was uploaded by Tobias Ide on 20 May 2015.
The user has requested enhancement of the downloaded file.
Why do conﬂicts over scarce renewable resources turn violent?
A qualitative comparative analysis
Tobias Ide *
Research Group Climate Change and Security, Institute of Geography, Cluster of Excellence Integrated Climate System Analysis and Prediction (CliSAP),
University of Hamburg, Grindelberg 7, 20144 Hamburg, Germany
1. Introduction
The possible impact of global environmental change on intrastate
violent conﬂict onset has received considerable attention
since at least the 1990s, with a special focus on the scarcity of
renewable natural resources (Ba¨chler, 1998; Homer-Dixon and
Blitt, 1998). In recent years, this research has been related to and
picked up by a growing literature on climate change and conﬂict
(e.g. Scheffran et al., 2012; Theisen et al., 2013). But the role of
renewable resources for violent conﬂict onset is not only discussed
by scientists, but by policy makers as well. US Secretary of State
John Kerry, for instance, recently expressed his concern about the
issue: ‘If we don’t respond adequately to the challenge of global
climate change over the course of these next years there will be
people ﬁghting wars over water and over land’ (U.S. Department of
State, 2013).
Research on the possible links between natural renewable
resources and conﬂict has been conducted in a range of disciplines,
including geography, political science, sociology and anthropology.
But despite considerable research efforts, no scientiﬁc consensus
on the issue has emerged as yet. Some quantitative studies suggest
a link between low precipitation levels (Fjelde and von Uexkull,
2012; Raleigh and Kniveton, 2012) or freshwater scarcity (Gizelis
and Wooden, 2010; Raleigh and Urdal, 2007) and intra-state
violent conﬂict, while others ﬁnd no signiﬁcant relationship
(O’Loughlin et al., 2014; Wischnath and Buhaug, 2014) or even a
negative correlation between low rainfall/water scarcity and
violent conﬂict within states (Hendrix and Glaser, 2007; Salehyan
and Hendrix, 2014). The same is true for quantitative studies on
soil degradation (Hendrix and Glaser, 2007; Raleigh and Urdal,
2007; Rowhani et al., 2011; Theisen, 2008) and deforestation (Esty
et al., 1999; Theisen, 2008). The ﬁndings of qualitative studies are
similarly ambivalent. Some authors claim a role for renewable
resource scarcity as a cause of violent conﬂict in certain cases
(Homer-Dixon, 1994; Kahl, 2006; Schilling et al., 2012), some
scholars reject such a link (Adano et al., 2012; Selby and Hoffmann,
2014), and some provide mixed results (Benjaminsen and Ba,
2009; de Chaˆtel, 2014).
In order to advance our knowledge about the links between
renewable resource scarcity and intra-state violent conﬂict,
scholars have repeatedly emphasized three tasks. First, according
to Barnett (2000), scholars have convincingly argued how resource
Global Environmental Change 33 (2015) 61–70
A R T I C L E I N F O
Article history:
Received 8 October 2014
Received in revised form 16 February 2015
Accepted 24 April 2015
Available online
Keywords:
Violence
Conﬂict
Resource scarcity
Environment
QCA
A B S T R A C T
This study addresses the question why intergroup conﬂicts over scarce, renewable resources in
peripheral areas of the global South escalate into violence. In order to do so, twenty cases of such
conﬂicts, seven of which turned violent, are analyzed. The method of fuzzy-set qualitative comparative
analysis is used in order to bridge the gap between quantitative and qualitative accounts in the ﬁeld and
to detect patterns of conjunctural causation. In theoretical terms, structural conditions (negative
othering and high power differences between the conﬂict parties) and triggering conditions (external
resource appropriation and recent political change) of a violent escalation of renewable resource
conﬂicts are distinguished. The empirical results as well as various robustness checks and comparisons
with individual cases suggest that the simultaneous presence of negative othering, low power
differences and recent political change is a sufﬁcient condition for the violent escalation of conﬂicts over
scarce renewable resources. I conclude that research on socio-environmental conﬂicts should pay more
attention to conjunctural causation, local power differences and qualitatively different forms of conﬂict
and political change.
ß 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
* Tel.: +49 040 428389196; fax: +49 040 428389211.
E-mail address: tobias.ide@zmaw.de
Contents lists available at ScienceDirect
Global Environmental Change
journal homepage: www.elsevier.com/locate/gloenvcha
http://dx.doi.org/10.1016/j.gloenvcha.2015.04.008
0959-3780/ß 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/
4.0/).
scarcity causes grievances, livelihood insecurities and conﬂicts, but
we do hardly know how, when and why such tense situations
escalate into open violence. Similarly, Engels/Chojnacki (2012: 94)
claim that ‘the transition from conﬂict to violence has not yet been
analyzed in a sufﬁciently sophisticated manner in the literature on
‘‘environmental conﬂicts’’.’
Second, the inconsistent results of previous studies suggest that
renewable resource scarcity is linked to violent conﬂict only if
speciﬁc (combinations of) context factors are present. This claim is
nowadays shared by nearly all authors in the research ﬁeld
(Homer-Dixon and Blitt, 1998; Scheffran et al., 2012). ‘It is
important to ask, therefore, why violence [related to scarce
resources] occurs in some places and not in others’ (Peluso and
Watts, 2001: 29). Recently, quantitative studies have tried to
address this problem by introducing interaction terms between
some independent variables, such as reduced precipitation,
political exclusion and economic marginalization (Fjelde and
von Uexkull, 2012; Theisen et al., 2012). However, the number of
interactions terms that can be used in a statistical regression is
limited (Vis, 2012). Case studies, by contrast, are able to consider
complex interactions between different variables, but often suffer
from a lack of generalizability and comparability.
This relates to a third, more general point. The methods most
widely used in the research on renewable resource scarcity and
violent conﬂicts are so far either large-N regression analyses or
qualitative single-case studies. The latter have repeatedly been
criticized for the low external validity of their ﬁndings (Gleditsch
and Urdal, 2002; Koubi et al., 2014), although case studies are in
principal able to produce generalizable results (Flyvbjerg, 2006).
But the shortcomings of large-N regressions have been pointed out
as well. Besides their limited ability to consider interaction terms
and non-linear effects (Sterzel et al., 2014), they cannot include
important variables on which quantitative datasets either do not
exist (e.g. resource distribution) or are hard to produce (e.g.
identities, traditional conﬂict resolution mechanisms) (Ide and
Scheffran, 2014; Selby, 2014). Other datasets have only a low
spatial resolution, usually the national level, or are criticized for
their low reliability (Ide and Scheffran, 2014; Koubi et al., 2014). As
a consequence, calls have been launched to explore middle ways
between qualitative single-case and quantitative large-N studies
which combine the strengths of both approaches (Meierding,
2013; Solow, 2013).
All these three suggestions are picked up by this study. In order
to do so, it utilizes the rich case-study literature on renewable
resource scarcity and conﬂict. Twenty cases of intergroup conﬂict
around land, water, ﬁsh or forest resources are identiﬁed, seven of
which escalated into open violence, while 13 remained largely
non-violent. The twenty cases are compared in a systematic
manner with the help of the qualitative comparative analysis
(QCA) technique/approach in order to detect the conditions under
which conﬂicts around scarce renewable resources turn violent.
QCA is well suited to deal with complex causal relationships and to
uncover relevant context factors and interaction effects. Since
cases are selected from a variety of locations and contexts, the
results are much more generalizable than single-case studies.
However, the analysis is still essentially based on the in-depth,
qualitative knowledge of the twenty cases under study. In this
sense, the QCA provides a middle ground between quantitative
large-N and qualitative case studies.
This article proceeds as follows. In the next section, the
theoretical background of the study is developed (2). Afterwards,
the research design (3) as well as the results and several robustness
checks (4) are presented. In the following discussion, the results
are interpreted and compared to in-depth analyses of singular
cases (5). Avenues for future research as well as policy implications
are spelled out in the conclusion (6).
2. Theoretical background
In this study, a conﬂict is deﬁned as a manifest clash of the
interests of two or more social groups. Violence refers to the use of
direct, physical force against human beings. And renewable
resource scarcity describes a situation in which the land, water,
ﬁsh or forest resources in a given area are insufﬁcient to satisfy
current human demands for these resources. Such scarcity can be
supply-induced, demand-induced, and/or structural (induced by
unequal distribution) (Homer-Dixon and Blitt, 1998: 6). But under
what conditions do conﬂicts around such scarce resources turn
violent? Albeit not without objections (Selby and Hoffmann, 2014),
previous research largely agrees that violent conﬂicts around
scarce renewable resources are most likely to occur in relatively
poor countries (often termed global south or non-OECD world),
and there especially in rural and peripheral areas where the state’s
capacity is limited (Bretthauer, 2014; Buhaug et al., 2010). These
factors are used to demarcate the ‘area of homogeneity’ of this
study, which makes sure that the cases selected are similar enough
to compare them in a meaningful way (Berg-Schlosser and de
Meur, 2009: 20f). But poverty and peripheral location are not
suitable for distinguishing cases of violent conﬂict from cases of
non-violent conﬂict about scarce renewable resources, since both
conditions are quite prevalent.
A starting point for my theoretical framework is the distinction
between structural and triggering conditions (roughly equal to
independent variables) of violent conﬂict escalation (Hendrix and
Glaser, 2007; Kaufman, 2001). The former are deﬁned as the preconditions
of a violent conﬂict which are largely static and
invariant over time, while the latter refer to short-term dynamics
or ‘precipitating events’ (Hislope, 2007: 154) of violent escalations.
In the QCA terminology, both structural and triggering conditions
are INUS conditions for the violent escalation of conﬂicts over
scarce renewable resources. An INUS-condition ‘is an insufﬁcient
but necessary part of a condition which is itself unnecessary but
sufﬁcient’ for the outcome under investigation (Schneider and
Wagemann, 2012: 79). Since the number of conditions that should
be included into a QCA is limited, the analysis will focus on four
conditions, which is in line with recent recommendations (BergSchlosser
and de Meur, 2009; Marx and Dusa, 2011). The
conditions are selected in accordance with the theoretical
literature, but also in a dialogue with the cases (Schneider and
Wagemann, 2012: 281). Only such conditions can be chosen on
which reliable and location-speciﬁc information are available for
all (potential) cases under study. More speciﬁcally, the analysis
will focus on two structural and two triggering conditions for the
violent escalation of conﬂicts over scarce renewable resources. It is
expected that a combination of structural and triggering conditions
is sufﬁcient for the violent escalation of conﬂicts over scarce
renewable resources, while the mere presence of (combinations of)
either structural or triggering conditions is neither necessary nor
sufﬁcient for such an escalation.
The ﬁrst structural condition used in the analysis is negative
othering. The importance of collective identities for the use of
violence by conﬂict parties (Fro¨hlich, 2012; Kaufman, 2006) as well
as the stability of such identities over time (Jabri, 1996) is well
known. Identities can be understood as collective social constructs
which deﬁne who are the members of a given social group, what
attributes and goals they share, and how they relate to other
groups (Abdelal et al., 2006). The delineation from other groups
(‘othering’) can facilitate the use of violence if the respective Other
is portrayed in negative terms (Chatterjee, 2012; Hansen, 2006). The
concrete forms ofnegative othering are time and placespeciﬁc.But it
has been shown that the construction of another group (a) as an
aggressor or existential threat to the Self and/or (b) as much lower in
value/legitimacy than the Self usually provides motivation and
T. Ide / Global Environmental Change 33 (2015) 61–7062
legitimacy for the use of violence in intergroup conﬂicts (e.g. Bar-Tal,
1998; Buzan et al., 1998; Kaufman, 2006). When conditions of
resource scarcity and perceived aggressive intentions by other
groups combine, they potentially produce a strong climate of
insecurity in which preliminary attacks to capture resources or
weaken the opponent are likely to happen (Kahl, 2006: 46f;
Scheffran et al., 2014). If other groups are conceived as having a low
status or legitimacy, the inhibition threshold to use violent means in
order to prevent these Others from using/acquiring renewable
resources recedes (Martin, 2005). Therefore, I expect that negative
othering is an INUS condition for the violent escalation of conﬂicts
over scarce renewable resources.
The second structural condition used is high power differences.
Power has been identiﬁed as an important yet understudied factor
in conﬂicts around scarce renewable resources (Houdret, 2012;
Selby and Hoffmann, 2014). In this context, Zeitoun (2008: 26–29)
distinguishes between three dimensions of power: hard power,
relational power, and ideational power. While the latter is crucial
for the occurrence or non-occurrence of a conﬂict and therefore of
minor relevance here, the ﬁrst two dimensions are important in
shaping the dynamics of conﬂicts around scarce resources. Hard
power is understood as ‘the material capacity of one party to gain
the compliance of the others’ (Zeitoun, 2008: 26) and includes
material wealth as well as the capability to apply direct, physical
violence. Relational power refers to the power one group derives
from its authority and legitimacy. Concrete manifestations of this
power include judicial backing, support from outside actors, media
attention, recognition by the international community or issue
linkage. But it is still unclear how (hard and relational) power
differences shape the escalation of conﬂicts. Research on
international wars has concluded that large and stable power
asymmetries reduce the risk of the violent escalation of interstate
conﬂicts, mainly because the weaker party is usually willing to
concede in order to avoid open confrontation (Bennett and Stam,
2004). In principle, this argument should hold true for intra-state
conﬂicts around renewable resources as well. On the other hand,
some studies suggest that large power differences increase the risk
for conﬂicts around natural resources to turn violent because the
stronger party has to fear neither considerable resistance nor
sanctions when applying physical force (Simmons et al., 2007;
Watts, 2004). Large power differences and the associated political
marginalization can also fuel grievances of the weaker party
(Raleigh, 2010). Consequentially, no directional expectation
regarding the relevance of large power differences as an INUS
condition can be made.
The ﬁrst triggering condition included in this study is external
resource appropriation, which is inspired by the political ecology
literature (e.g. Peluso and Watts, 2001; Robbins, 2004: 9–16).
Some authors argue that commercialization, understood as the
increasing exchange of natural resources on markets rather than
within local systems of reciprocity and subsistence, is an important
factor for the violent escalation of conﬂicts over scarce renewable
resources (Assies, 2003; Yeh, 2000). Other scholars highlight the
relevance of privatization, deﬁned as the transformation of a
resource from an open access or common pool to a private good, for
the violent escalation of resource conﬂicts (Jewitt, 2008; Simmons
et al., 2007). State interventions aimed at conservation or national
development, such as natural parks (Duffy, 2002) or state-led
water redistribution (Lynch, 2012), are claimed to have an
escalating effect as well. The hypothesized causal mechanisms
represented by the concepts of commercialization, privatization
and state intervention are very similar: Either are local groups
aggrieved since they are excluded from the use of scarce
renewable resources, for instance because of higher prices
(commercialization), an increasing scarcity due to resource
exports (commercialization) or the enclosure of resource-rich
areas (privatization, state intervention). Or powerful actors use
violent means in order to enforce property claims (privatization),
resource control (state intervention) or access to valuable resources
(commercialization). Therefore, and because only a limited number
of conditions can be used in a QCA, I follow the suggestion of
Schneider/Wagemann (2012: 277) to create a ‘master or macrovariable’.
External resource appropriation refers to a situation in
which commercialization and/or privatization and/or state intervention
are present. It can be deﬁned as a recent increase in the
appropriation of renewable resources by actors which are external
to the local conﬂict context (e.g. transnational companies, national
governments, urban consumers) and which do not use these
resources for subsistence purposes. The directional expectation is
that external resource appropriation is an INUS condition for the
violent escalation of conﬂicts over scarce renewable resources.
The second triggering condition used in the analysis is recent
political change. There is a large consensus about the importance of
regime change for the eruption of civil wars (Dixon, 2009). This
might be the case because regime change is already the result of a
severe (violent) conﬂict. But it is also possible that regime change
provides opportunities (e.g. low state capability to enforce law and
order) or motivations (e.g. resistance against the new regime’s
policies) for the use of violence. These arguments are also valid for
conﬂicts around scarce natural resources (Timura, 2001). However,
the focus on regime change alone might be too narrow. Changes of
major elements of the political system (which are not always
accompanied by regime change) can also have escalating effects on
conﬂicts about renewable resources. The re-drawing of administrative
boundaries, for instance, might cause the escalation of
conﬂicts around land or water resources in areas contested
between different groups (Kahl, 2006). Changes of policies related
to renewable resources, such as water privatization, can have
similar consequences (Assies, 2003). Therefore, political change is
deﬁned here as a modiﬁcation of (a) the characteristics of the
political system and/or its spatial organization, and/or (b) the laws
or policies related to the disputed renewable resources. However,
some forms of political change, such as the introduction of more
constitutional rights or the democratization of resource management,
might mitigate the use of violence in resource conﬂicts
(Ratner et al., 2013). Therefore, I do not formulate directional
expectations regarding recent political change as an INUS
condition for the violent escalation of conﬂicts over scarce
renewable resources.
3. Research design
3.1. QCA in a nutshell
QCA was developed in the 1980s in order to compare causal
patterns among a medium number of cases in a systematic way
(Ragin, 1987). More speciﬁcally, QCA aims at identifying necessary
and sufﬁcient conditions for a certain outcome. In order to do so, it
applies a set-theoretic perspective (Schneider and Wagemann,
2012), i.e. it analyzes whether given sets of cases stand in a subsetrelationship
or superset-relationship to each other. Regarding the
paper at hand, the question is then whether the set of cases in
which conﬂicts over scarce renewable resources turn violent is a
subset (indicating necessity) or a superset (indicating sufﬁciency)
of one of the (combinations of) conditions discussed in section 2, or
if no set relations can be detected (indicating the absence of causal
relationships).
When comparing cases, QCA obliges researchers to decide for
each case whether it belongs to the set of cases sharing a certain
condition (e.g. experiencing violence) or not (calibration). The
more recent variant of fuzzy-set qualitative comparative analysis
(fsQCA) allows researchers to go beyond simple in-out (1 or 0)
T. Ide / Global Environmental Change 33 (2015) 61–70 63
dichotomies and calibrate cases as having a partial membership
in a set (Ragin, 2009). However, the qualitative threshold of 0.5,
distinguishing between cases which are more in than out of a
certain set and vice versa, remains important (Schneider and
Wagemann, 2010). All cases can therefore be thought of as
combinations of different conditions, such as being a member of
set A (condition A is present), but not of B and C (conditions B
and C are not present). All possible combinations of conditions
are listed in a so-called truth table. Each of the cases
investigated can be assigned to one (and only one) combination
of conditions (called a truth table row), while often some
possible combinations (logical remainders) are not covered by
real cases. The number of truth table rows, and thus the number
of logical remainders, grows exponentially with the number of
conditions included. A high number of conditions in combination
with a low number of cases thus raises problems of limited
diversity (Ragin and Sonnett, 2005), theoretical interpretation
(Schneider and Wagemann, 2010: 402) and validity of the
results (Marx and Dusa, 2011).
Hug (2013) has recently argued that QCA scholars often do not
account for possible measurement errors during the calibration
process However, the fact that the calibration of the membership
scores is usually based on an in-depth knowledge of the cases
under consideration strongly reduces the number of measurement
errors (de Meur et al., 2009). The same is true for various
robustness checks of the results (see Skaaning, 2011 and Section 4).
Once the cases are calibrated, one can check which combinations of
conditions are linked to the outcome of interest, and these
combinations/truth table lines are minimized in order to achieve
the so-called solution formula or solution term. The solution
formula/term indicates the (combinations of) conditions which are
sufﬁcient for the outcome of interest (Ragin, 2009).
QCA is therefore based on three key assumptions (Schneider
and Wagemann, 2012: 78). Equiﬁnality means that the same
phenomenon can potentially be explained by different, mutually
non-exclusive sets of conditions. Conjunctural causation refers to
the possibility that a certain condition has no effect on the outcome
on its own, but only in combination with other conditions, for
instance in the form of INUS conditions. Both assumptions
resonate very well with the theoretical assumptions discussed
above as well as with recent efforts to introduce interaction effects
in the research on socio-environmental conﬂicts (Theisen et al.,
2012). Finally, causal asymmetry implies that the (combination of)
conditions causing a certain outcome (here: violent escalation) are
not necessarily a mirror image of those conditions causing the
absence of this outcome. This is in line with the widely shared
sentiment in peace and conﬂict research that violent conﬂict and
its absence cannot be treated as simple binary oppositions
(Chenoweth and Cunningham, 2013).
3.2. Case selection
The selection of cases is driven by several criteria. First, all cases
are supposed to be characterized by a conﬂict about scarce
renewable resources. Second, cases have to be part of the area of
homogeneity, so the conﬂicts should have taken place in
peripheral, rural areas of countries in the global south. Third,
since conﬂict patterns during the Cold War era might differ
signiﬁcantly from post-Cold War conﬂict dynamics, a temporal
dimension is added to the homogeneity space: The study is limited
to conﬂicts which took place after 1989. Similarly, only intra-state
conﬂicts are considered because violent inter-state conﬂicts about
renewable resources are very unlikely (Gleditsch, 2012). Fourth, in
order to increase the generalizability of the results through a
diverse case selection strategy (Gerring, 2007: 97–101), conﬂict
cases are chosen which
 took place during different time periods (between 1990 and
2010),
 took place in different geographical locations within the global
south (Asia, Africa, Latin America, Middle East, and the Paciﬁc),
 took place at various (sub-national) spatial scales (ranging from
village to federal state level), and
 involved different kinds of renewable resources (land, water,
ﬁsh, forest).
Fifth, the analysis only includes cases on which sufﬁcient and
reliable information is available. This implies that the cases are
discussed by several studies from different authors, that the
various studies and authors agree in their description of the cases,
and that at least parts of this literature should be peer-reviewed.
One should note that this prerequisite potentially reduces the
generalizability of the results due to research or publication biases.
Sixth, I consider only cases which are representative of their region.
In other words: If a conﬂict over scare renewable resources became
violent in one village, but not in the surrounding villages which are
characterized by similar conditions, then this conﬂict is not used in
the analysis.
When taking all six criteria into account, a sample of twenty
cases of conﬂicts over scarce renewable resources can be derived
from the rich qualitative literature on socio-environmental
conﬂicts. Fig. 1 summarizes the sample of these cases. Appendices
1 and 2 contain more information on all twenty conﬂicts,
including short descriptions of and a full list of references for
each case.
3.3. Calibration
As discussed above, it is important that the calibration process
is based on a good knowledge of the cases and on clearly deﬁned
thresholds between the different set membership scores (Schneider
and Wagemann, 2012: 32). In order to gain an in-depth
knowledge, desk-based studies of the twenty cases selected are
conducted. Existing studies on conﬂicts about scarce renewable
resources are used and complemented by more general publications
on the relevant countries/regions. The Polity IV index
(Marshall et al., 2012) is also consulted when membership scores
for recent political change are assigned. If I was unsure about a
speciﬁc calibration decision or if less information was available for
one case than for other cases, I contacted experts on the conﬂicts or
regions under investigation to provide additional insights (see
acknowledgements). Many cases were also discussed with
colleagues and student research assistants in order to increase
the reliability of the calibration process. In appendix II, I provide
short descriptions of all cases and justiﬁcations for the calibration
decisions. Appendix I gives information on the literature used for
the calibration of each case.
This study uses a four-value scheme for the calibration process.
Membership scores for the conditions and outcomes can be
calibrated as 0 (fully out of the set), 0.33 (more out than in), 0.67
(more in than out) and 1 (fully in). Such as scheme ‘is especially
useful in situations where researchers have a substantial amount
of information about cases, but the nature of the evidence is not
identical across cases’ (Ragin, 2009: 90). It allows for capturing the
diversity of social reality beyond 1–0 dichotomies, but does no
contain too many values which would make proper calibration
unreliable (Ragin, 2009). In the following section, information
about the qualitative anchors and the graded membership
deﬁnitions used in the calibration procedure is provided (Schneider
and Wagemann, 2012: 32, 277).
 If direct, physical violence was used frequently and in a
systematic manner by at least one conﬂict party and caused
T. Ide / Global Environmental Change 33 (2015) 61–7064
several fatalities, the case is calibrated as fully violent (1). A
membership score of 0.67 is attributed if at least one conﬂict
party used violence frequently, causing several fatalities, but the
use of violence was not systematic and the majority of conﬂict
actions were still non-violent. The ‘point of maximum indifference’
(Schneider and Wagemann, 2012: 32) is passed when
violence was used only sporadically and unsystematically, but
caused one fatality (0.33). A case is considered completely nonviolent
if no or hardly any physical violence was used and no
fatalities are reported (0).
 Negative othering: A case is considered as a full member of the
set if all or the large majority of groups involved perceived it’s
Others as an existential threat or as vastly inferior (1). When the
majority of conﬂict parties shared such perceptions, a set
membership score of 0.67 is assigned. The point of maximum
indifference is passed if only a minority of the parties perceived
its opponent(s) as an existential threat and/or strongly inferior
(0.33). If no parties involved in the conﬂict shared such
perceptions, the case is considered as fully out of the negative
othering set (0). In order to account for the possibility of reverse
causality (the use of violence causes negative othering), I
primarily use studies which trace the relevance of negative
othering for (and before) the onset (and violent escalation) of the
conﬂicts.
 High power differences: If one group had a considerable
advantage in terms of both hard and relational power, the case
is considered as a full member of the power differences set (1). If
one group commanded signiﬁcantly more hard or relational
power and the other groups were not able to compensate this by
advantages (or at least equivalence) regarding the other
dimension of power, the case is calibrated as 0.67. The point
of maximum indifference is crossed if one group was more
powerful with regard to one dimension of power, but this
advantage was largely ‘balanced’ by an advantage of the other
group or an equilibrium between both groups with regard to the
other dimension of power (0.33). If all parties involved in the
conﬂict had an equal amount of hard and relational power at
their disposal, the case is considered as fully out of the set (0).
 External resource appropriation: Full membership is assigned if
recent increases in external resource appropriation led to a
situation in which access of local groups to the renewable
resource was severely restricted or completely denied (1). If
considerable commercialization, privatization or state intervention
processes took place in the last (approximately) ﬁve years
and caused a situation in which a signiﬁcant amount of the
resource was appropriated by outsiders, in turn strongly
inﬂuencing the livelihoods of the groups in conﬂict, the case is
considered as more out than in (0.67). By contrast, if the degree of
external resource appropriation remained limited in spite of
recent increases, thus only marginally inﬂuencing local livelihood
patterns and strategies, the case receives a membership
score of 0.33. A case is calibrated as being fully out of the set
when no or hardly any external appropriation of the disputed
renewable resource occurred (0).
 Recent political change: The difference between minor and
major political changes is crucial in this context. A change is
considered minor if it affects secondary characteristics of the
political system (e.g. electoral thresholds) or the resource regime
(e.g. documentary procedures for obtaining land titles) and
major if it affects deﬁning characteristics of the political system
(e.g. shift from autocracy to democracy) or resource regime (e.g.
legalization of the sale of communal land). Full membership is
assigned if major changes in both the political system and the
regime concerning the disputed resource took place in the last
(approximately ﬁve) years (1). If either the political system or the
renewable resource regime experienced major changes, the case
is calibrated as 0.67. The point of maximum indifference is
passed when recent changes of the political system and/or the
renewable resource regime took place, but can be characterized
as minor (0.33). A case is considered to be fully out of the set if
neither its political system nor the relevant renewable resource
regime has changed in the last (approximately ﬁve) years.
Fig. 1. Sample of cases analyzed.
T. Ide / Global Environmental Change 33 (2015) 61–70 65
4. Results
In the ﬁrst step, it should be analyzed whether there is any
necessary condition for the outcome (Ragin, 2009). The most
important factor is assessing necessity in the consistency value,
which indicates the degree to which a condition leads to the
outcome across the cases. It is good practice to consider only those
conditions as necessary which have a consistency value of 0.9 or
higher (Schneider and Wagemann, 2012: 278). In this study,
neither the presence nor the absence of any of the four conditions
discussed in section 2 passes this threshold. The presence of
negative othering (0.83), the absence of high power differences
(0.83) and the presence of recent political change (0.79) have the
highest consistency values.
When applying the logical minimization procedure to the truth
table rows in order to detect sufﬁcient conditions, three solution
terms can be produced. The conservative solution is only based on
those truth table rows which correspond to empirically observed
cases. The intermediate solution draws on all empirically observed
truth table rows and those combinations of conditions which do
not correspond to empirical cases (logical remainders), but which
contribute to the parsimony of the solution terms and can be
assumed to produce the outcome of interest (here: violent
escalation). Finally, the solution based on all empirically observed
cases as well as all those truth table rows contributing to the
parsimony of the solution term is called the parsimonious solution.
The latter is usually considered as problematic since it can be based
on truth table rows which are neither empirically observed nor in
line with theoretical expectations (Schneider and Wagemann,
2012: 151–177). Proponents of fsQCA consider the intermediate
solution as superior (Ragin, 2009: 111), while other scholars
criticize the inclusion of non-observed cases and recommend
relying on the conservative solution (de Meur et al., 2009).
In this study, the logical minimization only includes truth table
rows backed by empirical cases which pass a raw consistency
threshold of 0.8 (Ragin, 2006). Five out of sixteen truth table rows
remain as logical remainders. When these are included in the
analysis, using the directional expectations spelled out in Section 2,
the intermediate and the conservative solution are the same. This
solution formula is depicted in Table 1 and can be read as follows:
The combination of negative othering, the absence of high power
differences between the conﬂict parties and either recent political
change or external resource appropriation is a sufﬁcient condition
for the violent escalation of conﬂicts over scarce renewable
resources. The solution is characterized by a high consistency value
(0.89) and can explain ﬁve out of seven cases of violent escalation.
Only Southern Para´ and Tonle Sap remain unexplained. In addition,
none of the thirteen cases without violent escalations is
characterized by this combination of conditions (although this
statement should be treated with caution due to QCA’s assumption
of causal asymmetry).
In recognition of QCA’s assumption of causal asymmetry
(Schneider and Wagemann, 2010), the absence of the outcome
is investigated. No necessary condition for conﬂicts over scarce,
renewable resources to remain non-violent can be found.
Regarding sufﬁciency, I use the directional expectations discussed
in Section 2. The resulting conservative solution is quite complex
and its theoretical interpretation would go beyond the scope of this
paper. The intermediate solution considers the absence of negative
othering in combination with either high power differences or
recent political change as sufﬁcient for a conﬂict about scare
renewable resource to be carried out non-violently (consistency:
0.9; coverage: 0.72).
Before the results are discussed in the subsequent section, their
robustness should be tested with a focus on the sufﬁcient
conditions for violent escalation. In general, fsQCA is considered
as a quite robust method due to the researcher’s deep knowledge of
the cases analyzed (de Meur et al., 2009) and because the
consistency values of the truth table rows are based on evidence
from all cases analyzed (Ragin, 2009: 119). Nevertheless,
Schneider/Wagemann (2012: 284–295) and Skaaning (2011)
emphasize the importance of checking the robustness of the
results vis-a`-vis changes in (a) the frequency thresholds, (b) the
consistency thresholds, (c) the cases analyzed, (d) the conditions
used, and (e) the calibration decisions. In an analysis with twenty
cases, any other frequency threshold than (at least) one (case per
truth table row) is hard to justify (Ragin, 2009: 107). The remainder
of this section is used to perform the other four robustness tests. If
not stated otherwise, a consistency threshold of 0.8 for the
inclusion of truth table rows into the minimization procedure is
used and the intermediate and conservative solutions are identical.
Firstly, the consistency threshold for the inclusion of truth table
rows in the minimization procedure (0.8 in the original analysis)
can be changed. A lowering of this threshold is not feasible since
the next lowest consistency score of a truth table row is beyond the
acceptable minimum threshold of 0.75 (Schneider and Wagemann,
2010). If the analysis is run again with a higher threshold of 0.85,
the solution formula looks very similar to the original solution
formula, but the absence rather than the presence of external
resource appropriation is identiﬁed as an INUS conditions for
violent escalation (Table 2, row 2).
Secondly, the robustness of the results vis-a`-vis a changing
population of cases should be checked. One might argue that
Guadalcanal differs from the other cases in which violence was
used because the conﬂict escalated into a civil war and a coup
d’e´tat (although due to the weak nature of the Solomon state,
the small territorial size of the country and the forms of
organized violence used, the dynamics of violence were similar
to those in other cases, such as Turkana or Oromiya). It could
also be claimed that in southern Para´, land was abundant,
maldistributed and hardly accessible (given the poor road
infrastructure) rather than scarce (although this argument
would be based on a limited understanding of resource scarcity).
Finally, one might recommend dropping the cases of Condega
and Dir Kohistan since no article on them is published in peerreviewed
journals (although sufﬁcient and reliable information
on both cases are available from edited volumes, working papers
and personal communication with experts). If I run the analysis
again but drop those four cases, the only difference to the
original solution formula is that external resource appropriation
is no longer included (Table 2, row 3). This change is driven by
the removal of Guadalcanal from the analysis, since it is the only
case of violent escalation uniquely covered by the causal
combination of negative othering, low power differences and
external resource appropriation.
Table 1
Conservative/intermediate solution for the violent escalation outcome.
Causal pathway negaoth * hipowdiff *
politchang
negaoth * hipowdiff *
extappro
Consistency 0.88 0.86
Raw coverage 0.58 0.50
Unique coverage 0.17 0.08
Cases covered Jabal Sabr, Northern
Nigeria, Southern
Oromiya, Turkana
Guadalcanal,
Southern Oromiya
Solution formula negaoth * hipowrdiff *
(politchang + extappro) ! violence
Solution consistency 0.89
Solution coverage 0.67
* = and.
+ = or.
 = absence of.
! = sufﬁcient for.
T. Ide / Global Environmental Change 33 (2015) 61–7066
Thirdly, a QCA with twenty cases can also be run with ﬁve
instead of four conditions (Marx and Dusa, 2011), although the
number of logical remainders grows with the inclusion of each
additional condition. Recent research has emphasized the importance
of education for the prevention of violence over scarce
renewable resources. Well-educated individuals can, for example,
introduce technologies to use scarce resources more efﬁcient or
seek employment in the secondary/tertiary sector (Bretthauer,
2014). Similarly, a low level of development is argued to facilitate
the violent escalation of conﬂicts over scarce renewable resources,
for instance because opportunity costs for joining a violent group
are lower (Barnett and Adger, 2007; Fjelde and von Uexkull, 2012).
I therefore run the analysis two additional times, once with high
levels of education (hiedu) and once with low levels of development
(lowdev) as an additional condition. The percentage of the
population which completed primary education is used as a proxy
for high education, while a high under-ﬁve mortality rate indicates
a low level of development. Sub-national (although not casespeciﬁc)
data on both indicators are obtained from the Demographic
and Health Survey Program (2013) (see appendix III for
more information on the calibration procedure, which has been
based on natural gaps in the data).
In accordance with theoretical expectations, the presence of
low levels of development and the absence of high levels of
education are identiﬁed as INUS conditions for the violent
escalation of conﬂicts over scarce renewable resources. But more
important is that negative othering, the absence of high power
differences and recent political change are again highlighted as
INUS conditions for violent escalation, while (the presence or
absence of) external resource appropriation is not part of the
solution formulas (Table 2, rows 4 and 5).
Fourthly, the scope for setting alternative calibration thresholds
is limited if the calibration procedure is based on deep qualitative
evaluations rather than on quantitative datasets. However, two
calibration-related robustness tests seem appropriate here. Firstly,
it might be argued that the recent political change condition mixes
up two too different concepts, namely changes of the political
system and changes of the resource regime. Therefore, the political
change condition is divided into two crisp-set (binary calibrated)
conditions: recent change of the political system (polsystchang)
and recent change of the resource regime (resregchang). If the
latter is used in the analysis, the solution formula identiﬁes
negative othering, low power differences and either external
resource appropriation or the absence of recent resource regime
change as sufﬁcient for a violent escalation (Table 2, row 6). This
contradicts the results of the previous speciﬁcations, which all
identify the presence (rather than the absence) of political change
as an INUS condition for violent escalation. When running the
analysis with the recent political change condition, a higher
consistency threshold for the inclusion of truth table rows (>0.8) is
chosen in order to produce a solution formula with a consistency
which is sufﬁciently high and comparable to those of the other
solution formulas. The combination of negative othering, low
power differences and recent changes of the political system
appears to be sufﬁcient for violent escalation (Table 2, row 7). This
result is in line with most other robustness tests. However, the
solution formulas based on the inclusion of either recent political
system change or recent change of the resource regime are not as
robust as the solution formulas identiﬁed when using the
condition of recent political change (results not shown).
Secondly, it can be argued that conﬂicts with one fatality
(calibrated as 0.33 in the analyses above) are more in than out of
the set of cases which experienced a violent escalation
(Sundberg and Melander, 2013). Therefore, the analysis is rerun
with a violent escalation outcome that is strictly calibrated
according to quantitative criteria: cases with zero fatalities are
calibrated as 0, cases with one fatality are calibrated as 0.67 and
cases with more than one fatality are calibrated as 1. The
resulting solution formula emphasizes the relevance of negative
othering, recent political change and especially low power
differences, but is ambivalent about the impact of external
resources appropriation (Table 2, row 8).
Where do these robustness checks with varying consistency
thresholds, cases included, conditions used and calibration
decisions leaves us? From Table 2, it can be seen that the
presence of negative othering and the absence of high power
differences are (parts of) INUS conditions for the violent escalation
of conﬂicts over scarce renewable resources in all eight solution
formulas found. The presence of recent political (system or
resource regime) change appeared as (a part of) an INUS condition
in seven of the eight analyses. The results regarding external
resource appropriation are ambivalent and in four robustness
tests, neither its presence nor its absence is identiﬁed as (part of)
an INUS condition. More importantly, the simultaneous presence
of negative othering, low power differences and recent political
change is a perfect superset of or identical to ﬁve of the eight
solution formulas produced, and a partial superset of seven of the
eight solution formulas. Hence the claim that this combination of
conditions is sufﬁcient for the violent escalation of conﬂicts over
scarce renewable resource is consistent and robust. Replication
data for all analysis presented in this section can be found in
appendix III.
5. Discussion
The causal pathway identiﬁed as consistent and robust in the
previous section (negaoth * hipowdiff * politchang) explains four
of the seven cases of violent escalation (Jabal Sabr, Northern Nigeria,
Southern Oromiya, Turkana) and is absent in all thirteen conﬂicts
which have not experienced a violent escalation. The violent
escalation of land conﬂicts on Guadalcanal in 1998 might also be
considered as covered by this causal pathway, because although the
Solomon Islands did not see a major change of the political system or
the land regime between 1993 and 1998, it experienced four
different governments, three governmental changes and tremendous
political volatility in this period (Moore, 2004).
Table 2
Original solution and results of the robustness tests.
Row Analysis Solution formula Cons. Cov.
1 Main analysis negaoth * hipowrdiff * (extappro + politchang) ! violence 0.89 0.67
2 Higher consistency threshold negaoth * hipowdiff * extappro * politchang ! violence 0.86 0.50
3 Four cases dropped negaoth * hipowdiff * politchang ! violence 0.87 0.77
4 Condition low development included negaoth * hipowdiff * politchang * lowdev ! violence 0.87 0.54
5 Condition high education included negaoth * hipowdiff * politchang * hiedu ! violence 0.88 0.58
6 Resource regime rather than political change negaoth * hipowdiff * (resregchang + extappro) ! violence 0.84 0.67
7 Political system rather than political change negaoth * hipowdiff * polsystchan ! violence 0.93 0.54
8 Violence threshold = one fatality hipowdiff * (extappro * politchang + negaoth * extappro) ! violence 0.85 0.59
9 Southern Para´ and Tonle Sap dropped negaoth * hipowdiff * (extappro + politchang) ! violence 0.88 0.79
T. Ide / Global Environmental Change 33 (2015) 61–70 67
In-depth investigations of the case of Southern Oromiya
(Ethiopia, 1995–2007) conﬁrm the causal mechanisms suggested
by the most robust solution formula. This vast, semi-arid region is
inhabited by various pastoralist ethnic groups (e.g. Borana, Digodi,
Guji) which frequently engage in (sometimes violent) conﬂicts
with each other. In the absence of lasting peace agreements, these
conﬂicts have not been resolved yet because the groups involved
are similarly powerful. Rather, continuing conﬂicts have been a
source of mistrust, fear and depreciation (negative othering). The
conﬂicts between the various pastoralist groups often concern
water, land and cattle resources, which became increasingly scarce
due to severe droughts in the 1980s. The replacement of the
communist Derg regime by a democratic and federalist political
system between 1991 and 1994 triggered an intensiﬁcation and
escalation of conﬂicts around land and water resources from 1995
onwards. Firstly, activities by the new government to commercialize
agriculture in Southern Oromiya increased resource scarcity
and access insecurities for pastoralists. Secondly, efforts by the
state to increase its presence in peripheral lowlands led to a
weakening of traditional conﬂict resolution mechanisms. And
ﬁnally, the attempt to ﬁx the boundaries between the territories of
various ethnic groups (‘ethnic federalism’) caused increasing
territorial conﬂicts between pastoral groups, particularly around
key water points and grazing areas (Hagmann and Mulugeta, 2008;
Temesgen, 2010).
However, the cases of Southern Para´ and Tonle Sap are not
covered by the most robust solution formula. Schneider and
Rohlﬁng (2013: 567f) discuss three causes of deviant cases
regarding coverage: a misspeciﬁed population, calibration errors
and omitted conditions. A misspeciﬁcation of the population is
supposedly of minor relevance here because Southern Para´ and
Tonle Sap do not belong to a set of extreme or speciﬁc cases of the
population studied. But a focus on calibration errors seems
promising. The most robust solution formula covers all ﬁve cases
in which the use of violence was largely two-sided, while it cannot
account for those two cases in which violence was primarily
exercised by one side. If the analysis is re-run without the cases of
Southern Para´ and Tonle Sap, the robust solution formula identiﬁed
above is largely conﬁrmed (Table 2, line 9). This suggests that the
outcome should be re-calibrated as ‘escalation into two-sided
violence’.
In order to detect omitted conditions, the conduction of an indepth
dissimilar-outcome comparison between the deviant cases
for coverage and individually irrelevant cases from the same truth
table rows is recommended (Schneider and Rohlﬁng, 2013). For the
analysis discussed here, this means comparing Southern Para´
(violent escalation despite of large power differences) to Dir
Kohistan (same truth table row but no violent escalation) and
Tonle Sap (violent escalation under conditions of high power
differences and low negative othering) to Galapa´gos and Koshi
Tappu (same truth table row but no violent escalation). In the latter
comparison, the issue of state strength appears to be salient. In the
case of Tonle Sap (Cambodia, 1993–2001), the country just
recovered from decades of intense (internationalized) civil war,
with ﬁghting activities still going on in some regions. Trust in state
institutions as well as the administrative and ﬁnancial capacities of
the state were also low (Kiernan, 2002; Seng et al., 2005). By
contrast, the conﬂict about land, forest and ﬁsh resources in Koshi
Tappu (Nepal, 1992–2001) took place before the intense phase of
the civil war. As in Gala´pagos (Ecuador, 1992–1999), state
institutions were effectively able to control the region and prevent
the widespread use of violence (Karki et al., 2006; Quiroga, 2009).
When comparing Southern Para´ (Brazil, 1990–1996) and Dir
Kohistan (Pakistan, 1993–2000), one might conclude that the
relevance of the contested resource for the livelihoods of the local
population is an important, but omitted variable. In Dir Kohistan,
forest products and timber royalities only complemented subsistence
agriculture as the main source of livelihood (Khan et al.,
2006). In Southern Para´, by contrast, highly unequal land
distribution and a lack of adequate agricultural plots posed a
strong threat to the livelihood of many local inhabitants since few
employment opportunities outside of subsistence agriculture were
available (Simmons et al., 2007). This is a plausible reason why
land conﬂicts in Southern Para´ escalated beyond the threshold of
systematic, deadly violence, while forest conﬂicts in Dir Kohistan
did not. Unfortunately, the operationalization of the concepts of
resource dependence (for livelihoods) and state strength remain
contested in the literature (Deligiannis, 2012; Newman, 2009).
Calibrating them for all twenty cases and re-running the analysis is
therefore beyond the scope of this study.
In theoretical terms, the most robust solution formula conﬁrms
the expectation that neither (combinations of) structural nor
(combinations of) triggering conditions by themselves are necessary
or sufﬁcient for the violent escalation of conﬂicts over scarce
renewable resources. This could be one reason why large-N studies
on resource scarcity and small-scale violence have so far not
produced consensual results (Selby, 2014). They are often based on
additive models that tend to miss a link between renewable
resource scarcity and violent conﬂict if causal relations are
conjunctural. The discussion of the cases not covered by the most
robust solution formula also suggest that different causal pathways
facilitate the escalation of renewable resource conﬂicts into
either one- or two-sided violence, while most of the literature
focuses on reciprocal violence (Scheffran et al., 2014: 371). In
addition, the results add further strength to the claim that a
stronger differentiation between causes of conﬂicts over renewable
resources (e.g. external resource appropriation) and causes of
the violent escalation of such conﬂicts (e.g. low power differences)
is necessary (Barnett, 2000). The ﬁnding that external resource
appropriation is not a robust INUS condition for violent escalation
contradicts parts of the political ecology literature (e.g. Allen,
2013; Jewitt, 2008).
The results indicate that lower power differences are robustly
identiﬁed as an INUS condition for violent escalation. By contrast,
many other studies hypothesize that conditions associated with
large power differences, such as exclusion, marginalization and
discrimination, facilitate the violent escalation of renewable
resource conﬂicts (Fjelde and von Uexkull, 2012; Kahl, 2006;
Reuveny, 2007). Recent political change also has been found to be a
triggering INUS condition for violent escalation, but has hardly
been considered in cross-case analyses of socio-environmental
conﬂicts. Finally, the relevance of negative othering as an INUS
conditions supports the growing, but still small number of
constructivist studies which argue for a stronger incorporation
of narrative and discursive factors in the research on socioenvironmental
conﬂicts (Fro¨hlich, 2012; Stetter et al., 2011;
Zeitoun et al., 2013).
6. Conclusion
Given the contradictory results previous studies report on the
issue, this study asked why some conﬂicts over scarce renewable
resource escalate into violence, while others do not. In order to
answer that question, twenty such conﬂicts, seven of which turned
violent, were analyzed. The method of fuzzy-set qualitative
comparative analysis (fsQCA) was employed in order to bridge
the gap between quantitative and qualitative approaches in the
research ﬁeld. My main result is that no single condition could be
identiﬁed as necessary or sufﬁcient, but that the simultaneous
presence of two structural conditions (negative othering and low
power differences) and one triggering condition (recent political
change) is sufﬁcient for the violent escalation of renewable
T. Ide / Global Environmental Change 33 (2015) 61–7068
resource conﬂicts. By contrast, high power differences (which are
implied by concepts like ethnic marginalization) and external
resource appropriation, which have been discussed as facilitating
conditions for the violent escalation of such conﬂicts (Fjelde and
von Uexkull, 2012; Jewitt, 2008), were not identiﬁed as robust
INUS conditions.
These ﬁndings open several prospects for future research on
renewable resource scarcity and violent conﬂict, an issue which
is currently mostly discussed in terms of climate change and
conﬂict. Firstly, it would be interesting to further test the most
robust solution formula through regression analyses and indepth
studies of additional cases. Especially relevant in this
context are cases which are characterized by the presence of both
structural conditions, but none of the two triggering conditions
for violent escalation. No such case was used in this analysis.
Secondly, my results suggest that the relationship between
resource scarcity and violent conﬂict is conjunctural, therefore
encouraging the use of interaction terms (Theisen et al., 2012)
and set-theoretic methods (Bretthauer, 2014) in future studies.
Thirdly, the factors causing conﬂicts over renewable resources
are not identical to those causing a violent escalation of such
conﬂicts. Subsequent analyses should account for this fact, for
instance by drawing on datasets for nonviolent conﬂicts (Day
et al., 2015). Fourthly, I have created several macro-variables for
this analysis. But the results of the robustness checks as well as
theoretical reasons suggest that a disaggregation of these
conditions might provide additional insights. This applies to
different forms of recent political change (political system vs.
resource regime change, democratization vs. autocratization)
and violence (one-sided vs. two-sided), but also to the various
dimensions of external resource appropriation (commercialization,
privatization, state intervention) and power differences
(hard vs. bargaining power). And ﬁfthly, state strength and the
importance of renewable resource for livelihoods were identiﬁed
as potential omitted conditions by dissimilar-outcome comparisons
of contradictory truth table rows. Future studies should
focus on these factors and dedicate more efforts to operationalize
and collect data on them.
When it comes to policy implications, this study clearly shows
that there is no automatic link between renewable resource
scarcity and violent conﬂict. Rather, this relationship is mediated
by three INUS conditions. Each of these conditions can be
addressed through policy measures, and if only one of them is
(signiﬁcantly) changed, the violent escalation of conﬂicts over
scarce renewable resources becomes unlikely. Since the prevention
of political change and the perpetuation or creation of large
power differences are at odds with democratic norms and human
rights, I would opt for reducing the amount of negative othering
between the competing groups. This can be done through, for
instance, discursive conﬂict transformation (Ramsbotham et al.,
2005: 288–301) or narrative conﬂict resolution (Buckley-Zistel,
2006).
Acknowledgements
Research for this study was funded in part by the German
Environmental Foundation (DBU) and the German Science
Foundation’s (DFG) Cluster of Excellence Integrated Climate
System Analysis and Prediction (CliSAP) (EXC177). I am grateful
to CliSAP and the School of Integrated Climate System Sciences
(SICSS) for making the open access publication of this article
possible. I thank Annabelle Houdret, Christopher Ward, Gerhard
Lichtentha¨ler, Giovanna Gioli, Helle Munk Ravnborg, Janpeter
Schilling, Lyla Mehta, Said Al-Shaybani, Shaheen Raﬁ Khan,
Talimand Khan, Tobias Haller, and Tor A. Benjaminsen for sharing
information on ‘their’ cases with me. Cosima Schro¨er and Sabrina
Wannewitz assisted me in collecting data on the cases. Pina von
Prozinski provided great help in designing Fig. 1. Anita Engels,
Giovanna Gioli and Ju¨rgen Scheffran provided helpful comments
on earlier versions of the paper. I am responsible for all remaining
errors.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.gloenvcha.2015.04.008.
References
Abdelal, R., Herrera, Y., Johnston, A.I., McDermott, R., 2006. Identity as a variable.
Perspect. Polit. 4, 695–711.
Adano, W., Dietz, T., Witsenburg, K.M., Zaal, F., 2012. Climate change, violent conﬂict
and local institutions in Kenya’s dryland. J. Peace Res. 49, 65–80.
Allen, M.G., 2013. Melanesia’s violent environments: towards a political ecology of
conﬂict in the western Paciﬁc. Geoforum 44, 152–161.
Assies, W., 2003. David versus Goliath in Cochabamba: water rights, neoliberalism,
and the revival of social protest in Bolivia. Latin Am. Perspect. 30, 14–36.
Ba¨chler, G., 1998. Why environmental transformations causes violence: a synthesis.
Environ. Change Secur. Project Rep. 4, 24–44.
Bar-Tal, D., 1998. Societal beliefs in times of intractable conﬂict: the Israeli case. Int.
J. Conﬂ. Manage. 9, 22–50.
Barnett, J., 2000. Destabilizing the environment-conﬂict thesis. Rev. Int. Stud. 26,
271–288.
Barnett, J., Adger, W.N., 2007. Climate change, human security and violent conﬂict.
Polit. Geogr. 26, 639–655.
Benjaminsen, T.A., Ba, B., 2009. Farmer-herder conﬂicts, pastoral marginalisation
and corruption: a case study from the inland Niger delta of Mali. Geogr. J. 175,
71–81.
Bennett, D.S., Stam, A.C., 2004. The Behavioral Origins of War. University of
Michigan Press, Ann Arbor.
Berg-Schlosser, D., de Meur, G., 2009. Comparative research design: case and
variable selection. In: Rihoux, B., Ragin, C.C. (Eds.), Conﬁgurational Comparative
Methods: Qualitative Comparative Analysis (QCA) and Related Techniques.
Sage, Thousand Oaks, pp. 19–32.
Bretthauer, J.M., 2014. Conditions for peace and conﬂict: applying a fuzzy-set
qualitative comparative analysis to cases of resource scarcity. J. Conﬂ. Resolut.,
http://dx.doi.org/10.1177/0022002713516841.
Buckley-Zistel, S., 2006. In-between war and peace: identities, boundaries, and
change after violent conﬂict. Millenium 35, 3–21.
Buhaug, H., Gleditsch, N.P.G., Theisen, O.M., 2010. Implications of climate change for
armed conﬂict. In: Mearns, R., Norton, A. (Eds.), Social Dimensions of Climate
Change: Equity and Vulnerability in a Warming World. World Bank, Washington,
DC, pp. 75–101.
Buzan, B., Wæver, O., de Wilde, J., 1998. Security: A New Framework for Analysis.
Lynne Rienner, Boulder.
Chatterjee, I., 2012. How are they othered? Globalisation, identity and violence in an
Indian city. Geogr. J. 178, 134–146.
Chenoweth, E., Cunningham, K.G., 2013. Understanding nonviolent resistance: an
introduction. J. Peace Res. 50, 271–276.
Day, J., Pinckney, J., Chenoweth, E., 2015. Collecting data on nonviolent action:
lessons learned and ways forward. J. Peace Res. 52, 129–133.
de Chaˆtel, F., 2014. The role of drought and climate change in the Syrian uprising:
untangling the triggers of the revolution. Middle East. Stud. 50, 521–535.
de Meur, G., Rihoux, B., Yamasaki, S., 2009. Addressing the critiques of QCA. In:
Rihoux, B., Ragin, C.C. (Eds.), Conﬁgurational Comparative Methods: Qualitative
Comparative Analysis (QCA) and Related Techniques. Sage, Thousand Oaks, pp.
147–165.
Deligiannis, T., 2012. The evolution of environment-conﬂict research: toward a
livelihood framework. Global Environ. Polit. 12, 78–100.
DHS Program, 2013. Demographic and Health Surveys, http://www.dhsprogram.
com/Publications/Publications-by-Country.cfm (21/03/2014).
Dixon, J., 2009. What causes civil war? Integrating quantitative research ﬁndings.
Int. Stud. Rev. 11, 707–735.
Duffy, R., 2002. Peace parks: the paradox of globalisation? Geopolitics 6, 1–26.
Engels, B., Chojnacki, S., 2012. Raus aus der Klimafalle: Wie die Friedens- und
Konﬂiktforschung mit o¨kologischem Wandel umgehen kann. J. Int. Peace Organ.
87, 89–110.
Esty, D., Goldstone, J., Gurr, T.R., Harff, B., Levy, M., Dabelko, G., Surko, P., Unge, A.,
1999. State failure task force report: phase II ﬁndings. Environ. Change Secur.
Proj. Rep. 5, 49–72.
Fjelde, H., von Uexkull, N., 2012. Climate triggers: rainfall anomalies, vulnerability
and communal conﬂict in Sub-Saharan Africa. Polit. Geogr. 31, 444–453.
Flyvbjerg, B., 2006. Five misunderstanding about case-study research. Qual. Inq. 12,
219–245.
Fro¨hlich, C., 2012. Security and discourse: the Israeli-Palestinian water conﬂict.
Conﬂ. Secur. Dev. 12, 123–148.
Gerring, J., 2007. Case Study Research: Principles and Practices. Cambridge University
Press, New York.
T. Ide / Global Environmental Change 33 (2015) 61–70 69
Gizelis, T.-I., Wooden, A.E., 2010. Water resources, institutions, and intrastate
conﬂict. Polit. Geogr. 29, 444–453.
Gleditsch, N.P., 2012. Whither the weather? Climate change and conﬂict. J. Peace
Res. 49, 3–9.
Gleditsch, N.P., Urdal, H., 2002. Ecoviolence? Links between population growth,
environmental scarcity and violent conﬂict in Thomas Homer-Dixon’s work. J.
Int. Aff. 56, 283–302.
Hagmann, T., Mulugeta, A., 2008. Pastoral conﬂict and state-building in the Ethiopian
lowland. Afr. Spectr. 43, 19–39.
Hansen, L., 2006. Security as a Practice: Discourse Analysis and the Bosnian War.
Routledge, London.
Hendrix, C.S., Glaser, S.M., 2007. Trends and triggers: climate, climate change and
civil conﬂict in Sub-Saharan Africa. Polit. Geogr. 26, 695–715.
Hislope, R., 2007. From expressive to actionable hatred: ethnic divisions and riots in
Macedonia. In: Jenkins, C., Gottlieb, E. (Eds.), Identity Conﬂicts: Can Violence be
Regulated? Transaction Publishers, New Brunswick/London, pp. 149–165.
Homer-Dixon, T., 1994. Environmental scarcities and violent conﬂict – evidence
from cases. Int. Secur. 19, 5–40.
Homer-Dixon, T., Blitt, J. (Eds.), 1998. Ecoviolence: Links Among Environment,
Population, and Security. Rowman & Littleﬁeld, Lanham.
Houdret, A., 2012. The water connection: irrigation, water grabbing and politics in
southern Marocco. Water Altern. 5, 284–303.
Hug, S., 2013. Qualitative comparative analysis: how inductive use and measurement
error lead to problematic inferences. Polit. Anal. 21, 252–265.
Ide, T., Scheffran, J., 2014. On climate, conﬂict and cumulation: suggestions for
integrative cumulation of knowledge in the research on climate change and
violent conﬂict. Global Change Peace Secur. 26, 263–279.
Jabri, V., 1996. Discourses on Violence: Conﬂict Analysis Reconsidered. Manchester
University Press, Manchester.
Jewitt, S., 2008. Political Ecology of Jharkhand Conﬂicts. Asia Pac. Viewp. 49, 68–82.
Kahl, C.H., 2006. States, Scarcity, and Civil Strife in the Developing World. Princeton
University Press, Princeton.
Karki, S., Upreti, B.R., Robinson, J., Bellbase, N., 2006. Koshi Tappu, Nepal. In: Moore,
P. (Ed.), Resource Rights, Sustainable Livelihoods, Environmental Security and
Conﬂict Mitigation in South Asia. IUCN, Bangkok, pp. 95–128.
Kaufman, S., 2001. Modern Hatreds: The Symbolic Politics of Ethnic War. Cornell
University Press, Ithaca/London.
Kaufman, S., 2006. Symbolic politics or rational choice? Testing theories of extreme
ethnic violence. Int. Secur. 30, 45–86.
Khan, S.R., Bokhari, S., Matthew, R., 2006. Livelihoods, security and conﬂict, Dir
Kohistan, Pakistan. In: Moore, P. (Ed.), Resource Rights, Sustainable Livelihoods,
Environmental Security and Conﬂict Mitigation in South Asia. IUCN, Bangkok,
pp. 129–159.
Kiernan, B., 2002. Conﬂict in Cambodia, 1945–2002. Crit. Asian Stud. 34, 483–495.
Koubi, V., Spilker, G., Bo¨hmelt, T., Bernauer, T., 2014. Do natural resources matter for
interstate or intrastate armed conﬂict? J. Peace Res. 51, 227–243.
Lynch, B.D., 2012. Vulnerabilities, competition and rights in a context of climate
change: toward equitable water governance in Peru’s Rio Santa Valley. Global
Environ. Change 22, 364–373.
Marshall, M.G., Jaggers, K., Gurr, T.R., 2012. Polity IV Annual Time-Series, 1800–
2011, http://www.systemicpeace.org/inscr/inscr.htm (16/04/2013).
Martin, A., 2005. Environmental conﬂict between refugee and host communities. J.
Peace Res. 42, 329–346.
Marx, A., Dusa, A., 2011. Crisp-set qualitative comparative analysis (csQCA): contradictions
and consistency benchmarks for model speciﬁcation. Methodol. Innov.
Online 6, 103–148.
Meierding, E., 2013. Climate change and conﬂict: avoiding small talk about the
weather. Int. Stud. Rev. 15, 185–203.
Moore, C., 2004. Happy Isles in Crisis: The Historical Causes for a Failing State in
Solomon Islands, 1998–2004. Asia Paciﬁc Press, Canberra.
Newman, E., 2009. Failed states and international order: constructing a post
Westphalian world. Contemp. Secur. Policy 30, 421–443.
O’Loughlin, J., Linke, A.M., Witmer, F.D.W., 2014. Effects of temperature and
precipitation variability on the risk of violence in sub-Saharan Africa, 1980–
2012. PNAS 111, 16712–16717.
Peluso, N.L., Watts, M., 2001. Violent environments. In: Peluso, N.L., Watts, M. (Eds.),
Violent Environments. Cornell University Press, Ithaca/London, pp. 3–38.
Quiroga, D., 2009. Crafting nature: the Galapagos and the making and unmaking of a
natural laboratory. J. Polit. Ecol. 16, 123–140.
Ragin, C.C., 1987. The Comparative Method: Moving Beyond Qualitative and
Quantitative Strategies. University of California Press, Berkeley.
Ragin, C.C., 2006. Set relations in social research: evaluating their consistency and
coverage. Polit. Anal. 14, 291–310.
Ragin, C.C., 2009. Qualitative comparative analysis using fuzzy sets (fsQCA). In:
Rihoux, B., Ragin, C.C. (Eds.), Conﬁgurational Comparative Methods: Qualitative
Comparative Analysis (QCA) and Related Techniques. Sage, Thousand Oaks, pp.
87–122.
Ragin, C.C., Sonnett, J., 2005. Between complexity and parsimony: limited diversity,
counterfactual cases, and comparative analysis. In: Kropp, S., Minkenberg, M.
(Eds.), Vergleichen in der Politikwissenschaft. VS, Wiesbaden, pp. 180–197.
Raleigh, C., 2010. Political marginalization, climate change, and conﬂict in African
Sahel states. Int. Stud. Rev. 12, 69–86.
Raleigh, C., Kniveton, D., 2012. Come rain or shine: an analysis of conﬂict and
climate variability in East Africa. J. Peace Res. 49, 51–64.
Raleigh, C., Urdal, H., 2007. Climate change, environmental degradation and armed
conﬂict. Polit. Geogr. 26, 674–694.
Ramsbotham, O., Woodhouse, T., Miall, H., 2005. Contemporary Conﬂict Resolution:
The Prevention, Management and Transformation of Deadly Conﬂicts. Polity
Press, Cambridge.
Ratner, B.D., Meinzen-Dick, R., May, C., Haglund, E., 2013. Resource conﬂict, collective
action, and resilience: an analytical framework. Int. J. Commons 7,
183–208.
Reuveny, R., 2007. Climate change-induced migration and violent conﬂict. Polit.
Geogr. 26, 656–673.
Robbins, P., 2004. Political Ecology: A Critical Introduction. Blackwell, Malden.
Rowhani, P., Degomme, O., Guha-Sapir, D., Lambin, E., 2011. Malnutrition and
conﬂict in East Africa: the impacts of resource variability on human security.
Clim. Change 105, 207–222.
Salehyan, I., Hendrix, C., 2014. Climate shocks and political violence. Global Environ.
Change 28, 239–250.
Scheffran, J., Brzoska, M., Kominek, J., Link, P.M., Schilling, J., 2012. Disentangling the
climate-conﬂict-nexus: empirical and theoretical assessment of vulnerabilities
and pathways. Rev. Eur. Stud. 4, 1–15.
Scheffran, J., Ide, T., Schilling, J., 2014. Violent climate or climate of violence?
Concepts and relations with focus on Kenya and Sudan. Int. J. Hum. Rights
18, 366–387.
Schilling, J., Opiyo, F., Scheffran, J., 2012. Raiding pastoral livelihoods: motives and
effects of violent conﬂict in north-eastern Kenya. Pastoralism 2, 1–16.
Schneider, C.Q., Rohlﬁng, I., 2013. Combining QCA and process tracing in settheoretic
multi-method research. Sociol. Methods Res. 42, 559–597.
Schneider, C.Q., Wagemann, C., 2010. Standards of good practice in
qualitative comparative analysis (QCA) and fuzzy-sets. Comp. Sociol. 9,
397–418.
Schneider, C.Q., Wagemann, C., 2012. Set-theoretic Methods for the Social Sciences:
A Guide to Qualitative Comparative Analysis. Cambridge University Press,
Cambridge.
Selby, J., 2014. Positivist climate conﬂict research: a critique. Geopolitics 19, 829–
856.
Selby, J., Hoffmann, C., 2014. Beyond scarcity: rethinking water, climate change and
conﬂict in the Sudans. Global Environ. Change 29, 360–370.
Seng, K., Song, S.L., Navy, H., Leang, S., 2005. Conﬂicts Arising from Re-allocation of
Fishing Lots: Perceptions from Community Fisheries in Cambodia. World Fish
Center, Phnom Penh.
Simmons, C., Walker, R.T., Arima, E.Y., Aldrich, S.P., Caldas, M.M., 2007. The Amazon
land war in the south of Para´. Ann. Assoc. Am. Geogr. 97, 567–592.
Skaaning, S.-E., 2011. Assessing the robustness of crisp-set and fuzzy-set QCA
results. Sociol. Methods Res. 40, 391–408.
Solow, A.R., 2013. A call for peace on climate and conﬂict. Nature 497, 179–180.
Sterzel, T., Lu¨ deke, M., Kok, M., Walther, C., Sietz, D., de Soysa, I., Lucas, P., Janssen, P.,
2014. Armed conﬂict distribution in global drylands through the lens of a
typology of socio-environmental vulnerability. Reg. Environ. Change 14,
1419–1435.
Stetter, S., Herschinger, E., Teichler, T., Albert, M., 2011. Conﬂicts about water:
securitization in a global context. Conﬂ. Coop. 46, 441–459.
Sundberg, R., Melander, E., 2013. Introducing the UCDP georeferenced event dataset.
J. Peace Res. 50, 523–532.
Temesgen, A.K., 2010. Climate Change to Conﬂict? Lessons from Southern Ethiopia
and Northern Kenya. Fafo, Oslo.
Theisen, O.M., 2008. Blood and soil? Resource scarcity and internal armed conﬂict
revisited. J. Peace Res. 45, 801–818.
Theisen, O.M., Gleditsch, N.P., Buhaug, H., 2013. Is climate change a driver of armed
conﬂict? Clim. Change 117, 613–625.
Theisen, O.M., Holtermann, H., Buhaug, H., 2012. Climate wars? Assessing the claim
that drought breeds conﬂict. Int. Secur. 36, 79–106.
Timura, C., 2001. Environmental conﬂict and the social life of environmental
security discourse. Anthropol. Q. 74, 104–113.
U.S. Department of State, 2013. Secretary Kerry holds a Google+ hangout with NBC’s
Andrea Mitchell, http://www.state.gov/secretary/remarks/2013/05/209273.
htm (15/11/2013).
Vis, B., 2012. The comparative advantages of fsQCA and regression analysis for
moderately large-N analyses. Sociol. Methods Res. 41, 168–198.
Watts, M., 2004. Resource curse? Governmentality, oil and power in the Niger delta,
Nigeria. Geopolitics 9, 50–80.
Wischnath, G., Buhaug, H., 2014. On climate variability and civil war in Asia. Clim.
Change 122, 709–721.
Yeh, E.T., 2000. Forest claims, conﬂicts and commodiﬁcation: the political ecology of
Tibetan mushroom-harvesting villages in Yunnan province, China. China Q. 161,
264–278.
Zeitoun, M., 2008. Power and Water in the Middle East: The Hidden Politics of the
Palestinian-Israeli Water Conﬂict. Tauris, London.
Zeitoun, M., Talhami, M., Eid-Sabbagh, K., 2013. The inﬂuence of narratives on
negotiations and resolution of the upper Jordan river conﬂict. Int. Negot. 18,
293–322.
T. Ide / Global Environmental Change 33 (2015) 61–7070
View publication statsView publication stats