Abstract
This study applies fuzzy-set Qualitative Comparative Analysis (fsQCA) to the debate on links between resource scarcity and armed conflict. Previous studies on this relationship have reached contradictory results. This study aims to solve this contradiction by arguing that social, economic, and political conditions play an important role in determining whether armed conflict erupts over resource scarcity. I test three theoretic hypotheses, focusing on weak states, economic situations of households, and human ingenuity. I compare fifteen resource scarce cases with conflict to sixteen cases without armed conflict. My analysis supports the hypothesis that the economic situation of households and the levels of human ingenuity matter. In particular, the impact of high dependence on agriculture and low levels of tertiary education on the link between resource scarcity and conflict is discussed. While employing an fsQCA proves a valuable step in accounting for contradictory results, limits of the methods are apparent as well.
Keywords
Since its inception in the late 1980s, qualitative comparative analysis (QCA; Ragin 1987) has been applied to a growing number of social science fields. While there are some application in the field of conflict studies (e.g., Pinfari 2011; Van der Maat 2011), it has not yet become a common method. This article explores the usefulness of QCA as a method by applying it to the issue of resource scarcity and its impact on armed conflict.
While the idea that climate change and resource scarcity can lead to conflict has become a widespread assumption in the media and policy circles, the picture drawn by academic research is more blurry. There are strong theoretic arguments advancing the link between resource scarcity and armed conflict, and this link is supported by a number of empirical studies. However, empirical evidence rejecting this link also exists. The aim of this study is to explain these contradictory results by focusing on the economic, political, and social conditions within these countries. I argue that the causal mechanisms that most empirical studies focus on only take place under very specific conditions.
Resource Scarcity and Armed Conflict
The debate on the links between resource scarcity and armed conflict was shaped in the 1990s by the research group around Thomas Homer-Dixon (1999) at the University of Toronto and the Environmental Change and Security Project at the Swiss Peace Foundation (Bächler 1998). Both research groups focused on exploring the causal mechanisms between resource scarcity and armed conflict through in-depth case studies. Their key argument has been put like this: “so great are the stresses generated by too many people making too many demands on their natural-resource stocks and their institutional support systems, that the pressures often create first-rate breeding grounds for conflict” (Myers 1987, 16). Much focus of resource scarcity theorists has been on the causal mechanisms that link resource scarcity to armed conflict. The proposed strong link between resource scarcity and armed conflict has drawn much criticism from political ecologists, neoclassical economists and from a cornucopian perspective.
Political ecologists reject the strong causal link drawn by resource scarcity theorists and argue that links between the environment and violent conflict are manifold and complex and that taking economic and social structures and the local context into account is paramount (e.g., Peluso and Watts 2001). In particular, the political and social nature of the land (such as the entitlements to land, its distribution, and the actors involved) play an important role in mediating human–nature interrelations. The neo-Malthusian view of resource scarcity theorists is therefore criticized as overly deterministic.
Neoclassical and Cornucopian scholars argue that conflicts over resources can be prevented by overcoming scarcity. The basic neoclassical argument states that market mechanisms mitigate scarcity through prices and substitution (Lomborg 2001, 148). Cornucopians believe that human inventiveness and technological advances can overcome resource scarcities, for instance through the increase in agricultural production.
Empirical evidence in the literature on the links between resource scarcity and armed conflict is mixed. The link is supported by a number of case studies, focusing on a wide range of countries, including, for instance, Rwanda (André and Platteau 1998), Kenya (Kahl 2008), the Philippines (Kahl 2008), and Bangladesh (Swain 1996). However, there are also case studies pointing to other factors besides resource scarcity such as grievances like corrupt officials or the marginalization of groups and their lifestyle (e.g., Turner 2004; Benjaminsen 2008; Benjaminsen et al. 2012).
Quantitative studies do not provide conclusive results on the links between various indicators of resource scarcity and armed conflict either. While most studies find no or weak results, there are also some studies that support the link between scarcity and conflict. The strongest statistical support for resource scarcity theories comes from an early study by Hauge and Ellingsen (1998) who established strong relationships between land degradation, freshwater scarcity, population density, deforestation, and conflict but results could not be replicated (Theisen 2008). Raleigh and Urdal (2007) find weak or insignificant effects of water scarcity and land degradation on armed conflict. The State Failure Taskforce (Esty et al. 1998) found no correlation between indicators of environmental scarcity such as soil degradation and deforestation rate and state failure. Studies focusing on demographic pressure found no links to conflict (Urdal 2005; Collier and Hoeffler 2004; Buhaug and Rød 2006). An exception is a study by Raleigh and Urdal (2007) that finds a positive relationship between population growth and density and violent conflict. The spotlight on the social consequences of climate change has sparked a number of studies focusing on links between climate and armed conflict but results have been inconclusive as well (for a recent review, see Gleditsch 2012).
The generally conflicting results point to a more complex relationship between scarcity and conflict than most resource scarcity theorists envision. In this study, I follow the critique of political ecologists in arguing that the context-specific conditions matter. However, in contrast to most political ecologists, I aim to compare various cases to find patterns of conditions across cases. I am choosing this approach to shed light on the conflicting results of empirical studies. In difference to both quantitative and qualitative empirical studies, I am not focusing on the question if resource scarcity is linked to conflict, but under which social, economic, and political conditions conflict occurs in resource-scarce countries. From this question follows a research design, which differs substantially from a statistical approach. While a statistical analysis would typically include a measure of both resource scarcity and conflict, the focus in this article is on how resource-scarce countries experiencing conflicts differ in their social, economic, and political conditions from resource-scarce countries that remain peaceful. Therefore, the analysis is limited to countries experiencing resource scarcity, making resource scarcity a scope condition rather than an independent variable. Consequently, the results do not answer the question whether resource scarcity leads to conflict but shed light on the social, economic, and political conditions that favor the outbreak of conflict.
Social, Economic, and Political Conditions
This study argues that conditions in resource-scarce countries can explain when conflicts erupt and when they do not. Most causal mechanisms proposed by resource scarcity theorists are indirect via different social, political, or economic intermediate factors. I am testing three hypotheses based on theoretical arguments explaining how resource scarcity leads to armed conflict in case studies based on state weakness, on the economic situation on a household level and on human ingenuity.
Weak States
Explanations of conflict are often centered on state weakness and situations where the state “does not have the resources to create legitimacy by providing security and other services. In its attempt to find strength, it adopts predatory and kleptocratic practices or plays upon and exacerbates tensions between the myriads of communities that make up the state” (Holsti 1996, 117). State weakness is an important intermediate factor in many accounts of resource scarcity theorists. Increasing problems due to mounting environmental scarcity lead to higher demands on the state, including those for development projects to mitigate the social effects of degradation and resource loss as well as social demands resulting from rural-to-urban migration, such as housing, employment, transport, and energy. The widening gap between demands placed on and revenues available to the state breeds discontent (Homer-Dixon 1994, 25).
One state-centered mechanism, termed resource capture by Homer-Dixon (1999, 74) and state exploitation by Kahl (2008, 50-51) focuses on the role of elites. In a situation of scarce resources, elites are tempted to use their power to secure their access to (potentially) scarce resources. By manipulating state policies in their favor, they can limit access to resources to themselves and their supporters, in particular those belonging to their own ethnic group. This weakens the institutions of the state and may contribute to social unrest or even spark conflict. Thus, weak political institutions are combined with systems of patronage and ethnic exclusion in an effort to exploit the state.
In many case studies of conflicts in resource scarce societies, corruption or patronage plays an important role including Kenya (Kahl 2008; Boone 2011), Rwanda (André and Platteau 1998), the Philippines (Kahl 2008), Mali (Benjaminsen and Ba 2009), and Tanzania (Benjaminsen, Maganga, and Abdallah 2009).
These systems of patronage are often closely linked to the exclusion of ethnic groups. In corrupt systems, issues concerning mainly disadvantaged groups are unlikely to be addressed, leading to grievances. In case studies of resource scarce societies, ethnic heterogeneity is often emphasized, but rarely seen as a key factor. In situations of resource capture, these three conditions interact.
Economic Situations of Households
The economic situation of households such as high dependence on subsistence agriculture and high poverty levels are emphasized in many case studies in resource scarce societies such as the Philippines (Kahl 2008), Somalia (Farah, Hussein, and Lind 2002), and Burundi (Oketch and Polzer 2002), but are discussed less in the theoretical mechanisms envisioned by resource scarcity theorists, who more often focus on the role of the state (but see de Soysa and Gleditsch 1999; Buhaug, Gleditsch, and Theisen 2008).
In societies where a large percentage of the population depend on agriculture for their livelihoods, resource scarcity threatens the livelihoods of a larger number of people than in societies where only very few people live off their land. For instance, Oketch and Polzer (2002, 90) argue that the combination of environmental pressure and lack of economic alternatives to subsistence agriculture ultimately leads to violence. As agriculture is often linked to high levels of poverty, it has been argued that poverty provides the link between dependence on agriculture and conflict (de Soysa and Gleditsch 1999).
In a comparative study of the impact of resource scarcity on violent conflict in Uganda, Rwanda, Ethiopia, and Burundi, Ejigu (2009, 889) argues that the close link between the environment and poverty has an impact on the relationship between resource scarcity and conflict. As the natural resource base is shrinking, there are few routes to escape poverty, increasing the risk for conflict to break out over resources. Poverty is also seen as an important factor in economic theories of civil war, which often focus on the opportunity costs of civil war, arguing that in countries with a higher per capita income people have more to lose in terms of income when they join a rebel group (Collier and Hoeffler 2004).
Human Ingenuity
Ingenuity are the “ideas applied to solve practical technical and social problems” (Homer-Dixon 1999, 109), which is an argument most prominently advanced by the Cornucopian perspective on scarcity. Ingenuity is differentiated into social and technical ingenuity. The argument is that resource-scarce countries need both, but that social ingenuity (or human capital) is a precursor for technical ingenuity as resource-scarce countries first need sophisticated and stable systems of markets, legal regimes, financial agencies and educational and research institutions to promote the development and distribution of new grains adapted for dry climates and eroded soils, of alternative cooking technologies to compensate for the loss of firewood, and of water conservation technologies. (Homer-Dixon 1999, 110)
Social ingenuity allows societies to adapt to consequences of environmental change in ways that go beyond technological innovation, for instance by establishing emergency credits in cases of crop failure or by anticipating food shortfalls and effectively organizing food aid (Homer-Dixon 1999, 111).
Human inventiveness provides a way to overcome problems generated by resource scarcity in a nonviolent way. This differs from other conditions mentioned, as it is a strategy for the mitigation of conflict rather than a contributing factor. While the absence of ingenuity does not lead to conflict directly, countries that experience conflict lack ingenuity as a mitigating factor.
Methods and Case Selection
These three hypotheses—weak states, economic situations of households, and human ingenuity—are tested by comparing thirty-one countries, which are shown to be resource scarce in terms of freshwater and arable land. I employ a set-theoretic approach of QCA to capture the conjunctural causation between resource scarcity and six conditions to explain previous contradictory empirical results.
QCA
QCA is a set-theoretic approach developed by Ragin (1987, 2008) that focuses on the presence and nonpresence of certain conditions. If interested in three conditions A, B, and C for the outcome Y, a number of possible combinations exist: A can be present, but neither B nor C; A and B can be present, but not C; neither A, B or C can be present and so on. QCA employs a table of all possible combinations of conditions and then allocates the empirical cases to these combinations. It is within this process that the qualitative nature of QCA comes into play as in-depth knowledge of cases is needed to calibrate conditions. In the next step, these sets of conditions are minimized to provide a solution formula.
Epistemologically, QCA is based on multiple conjunctural causality allowing for, for instance, different pathways to lead to the same outcome and for one condition to have different effects in different contexts. QCA is particularly beneficial to the analysis of context conditions as it focuses on the pathways and combinations of conditions that lead to an outcome. In particular, the case study literature on resource scarcity and conflict often uses conjunctural rather than additive causation. For instance, Bächler (1998, 25) argues that environmental degradation leads to conflict if “social fault lines can be manipulated in struggles over social, ethnic, political, or international power,” in fact arguing that environmental degradation leads to conflict only in conjunction with a number of other conditions. Using QCA in an effort to explain contradictory results on the links between resource scarcity and conflict has been suggested earlier (Theisen 2008) but has not received much attention so far.
Case Selection
When is an area resource scarce? Various measures such as population density (Hauge and Ellingsen 1998; Buhaug and Rød 2006; Theisen 2008) or the degree of soil erosion (Esty et al. 1998; Hauge and Ellingsen 1998; Hendrix and Glaser 2007; Raleigh and Urdal 2007) have been used to determine whether an area is resource scarce, but no standard operationalization has been established. I define resource scarcity in terms of arable land and freshwater as the most relevant resources for sustaining human livelihoods. To determine land-scarce cases, I use an index of land resource potential and constraints published by the UN Food and Agricultural Organisation (FAO 2000). I operationalize countries 1 as resource scarce where the equivalent potential arable land is less than .1 ha per capita. Equivalent potential arable land adjusts the land available by a number of factors that have an impact on land potential, such as soil characteristics, terrain characteristics, climate regime, and different crop types. The threshold of .1 ha builds on a study by Smil (1993, 69) that establishes .07 ha per person as the necessary threshold for a person to sustain their livelihood.
To determine water-scarce cases, I use the Falkenmark indicator, which is based on assessing the gap between the water needed to satisfy an individual’s needs and the water available to that person. Falkenmark, Lundqvist, and Widstrand (1989) establish three thresholds of renewable water resources per capita per year, based on water needs in the household, the agricultural and energy sectors as well as the needs of the environment. A country with less than 1,700 m3 of renewable water resources per capita per year is considered water stressed; one with less than 1,000 m3, water scarce; and if countries have less than 500 m3, they experience absolute water scarcity. I set the threshold to 1,000 m3 per capita per year and use data on total renewable freshwater resources per year and person from the FAO Aquastat main country database (FAO 2011) and the average value between 1990 and 2010.
Of the resulting thirty-nine cases, twenty-three have experienced conflicts between 1990 and 2010 according to Peace Research Institute Oslo (PRIO)/Uppsala Conflict Data Programs’ (UCDP 2011a) definition of conflict, using a threshold of twenty-five battle deaths. I limit my research to intrastate conflict (with and without foreign involvement) and nonstate conflicts, thus excluding conflicts between countries as most conflicts since World War II fall into the former categories and wars between states tend to exhibit different patterns from wars within states. I use the UCDP’s data sets on armed conflict (UCDP 2011b) and on nonstate conflict (UCDP 2011c).
In order to reduce the impact of reverse causalities, such as conflict leading to environmental degradation and resource scarcity, I have dropped cases that experienced armed intrastate conflict between 1980 and 1990. Cases in which an armed conflict was ongoing in 1990 are excluded, except for cases that did not reach the battle death threshold between 1980 and 1990. The case selection is summarized in Table 1.
Case Selection Including Indicators of Freshwater and Land Scarcity and Conflict Dates.
Note: aBold typeface denotes values that fall into my definition of resource scarcity (a value of less than 1,000 on the Falkenmark indicator or 0.1 or less equivalent potential arable land).
bCountries where data was not available are denoted by (—).
I use an fsQCA, where conditions and outcome are calibrated into membership scores between zero and one, with cases coded zero considered fully out of the set, cases coded one fully in the set. Cases that are more out of than in the set have a score lower than 0.5 and cases that are more in than out of the set have a score higher than .5. Thus, .5 is the score with the highest ambiguity where cases are neither in nor out of the set. To calibrate outcomes and conditions, I am using the calibrate function in the fsQCA software (Davey and Ragin 2009), which requires the researcher to set thresholds for being fully in the set (fuzzy-set score of 1), being fully out of the set (fuzzy-set score of 0) and for the crossover point (fuzzy-set score of .5). Thresholds need to make theoretic sense and the cases need to be distributed across the spectrum.
My main outcome is in terms of total battle deaths (UCDP 2011d). I have set the crossover threshold at twenty-five battle deaths, which is the threshold for cases to be considered a conflict case and the fully in threshold at 1,000, which is the threshold for an armed conflict to be considered a civil war. The threshold for being fully out of the set is set at no battle-related deaths.
Conditions
The three hypotheses include six conditions: quality of political institutions, corruption levels, and ethnic exclusion in the state weakness hypothesis; poverty levels and dependence on agriculture in the economic situation of households hypothesis; and tertiary education in the human ingenuity hypothesis.
In an effort to avoid reverse causalities, such as conflict leading to poverty or weak political institutions, data for cases that experienced conflict are taken from the year prior to the conflict. If these data were not available, the closest available data were used. For nonconflict cases, the data were averaged out over the twenty-one-year period.
Quality of Political Instituions (PIN)
To measure the quality of political institutions, I am using the civil liberties index of Freedom House Index, which includes a measurement of the rule of law to measure quality of political institutions. While the Freedom House Index has met a lot of criticism (Munck and Verkuilen 2002), there are, unfortunately, very few indicators that span the entire time frame and a global sample. Freedom House categorizes countries as “free” where their score is between 1 and 2.5, as partly free between 3 and 5 and as not free between 5.5 and 7. According to this, I have used 5.5 as threshold for fully in the condition of state weakness and 2.5 as threshold for full out. For the crossover threshold, I avoid using the 4, as this would leave seven cases coded as neither in nor out. Instead, the crossover point is set at 4.1.
Political Corruption (COR)
I employ the World Bank’s Control of Corruption indicator, which captures “perceptions of the extent to which public power is exercised for private gain” (World Bank 2012). Countries are scored between −2.5 and 2.5 with higher values representing less corruption. I keep the crossover point of 0, as it is established in the data set and then calibrate countries as fully within the set of countries experiencing corruption if the corruption index is −1.0 or lower and as fully out of the set of corrupt countries if it is 1.0 or higher.
Ethnic Exclusion (EEX)
Ethnic exclusion is operationalized as the percentage of the population excluded from power following Wimmer, Cederman, and Min (2009). As no clear thresholds are given within the data set, I concluded to use 20 percent as a threshold for fully in the data set of ethnically excluded groups as this is a naturally occurring gap in the data. The threshold for being more in than out of the sample of countries with excluded groups is set fairly low at 5 percent. As even fairly small excluded groups can have quite an impact, this makes theoretic sense. In addition, it utilizes another (smaller) gap in the data. Finally, the threshold for fully out of the set of countries that exclude parts of their population is set at 0 percent, resulting in thirteen countries where no groups of the population are excluded from the political game.
Poverty (POV)
Poverty is operationalized in absolute terms using the World Bank’s poverty headcount ratio at $1.25 a day. Basing thresholds on outside definitions is difficult in this case as it is in the nature of poverty indicators and threshold to be targeted toward policy goals. Therefore, the thresholds in this study are based on natural gaps occurring in the data. Eight countries in the sample have no people living on less than $1.25 a day, so this is taken as the first threshold. The crossover point is set at 10 percent between Algeria (7.6 percent) and Armenia (11.3 percent). Finally, countries are considered fully in the sample of countries with high poverty rates if the rate is more than 30 percent, dividing Mongolia (22.7 percent) and Uzbekistan (31.8 percent).
Dependence on Agriculture (DEP)
To measure dependence on agriculture, I use data on the agricultural population as percentage of the total population, defined as all persons depending for their livelihood on agriculture, hunting, fishing, and forestry, provided by the FAO (2009). To establish thresholds, I follow Alexandratos (1999) who defines a high dependence on agriculture as 50 to 80 percent of the population depending on agriculture as the main source of living. I use 50 percent as the threshold for countries to be more in than out of the set of countries with high levels of dependence on agriculture and 80 percent as the threshold for countries to be fully in the set. As a threshold for fully out of the set, I use less than 5 percent of the population depending on agriculture.
Tertiary Education (EDU)
The supply of ingenuity is determined by a number of factors, including the availability of financial and intellectual capital, the capacity to generate practical knowledge and the willingness of society to undergo social and technological change (Homer-Dixon 1999, 111). To measure human ingenuity, I focus on the intellectual capital, as this is most closely linked to “society’s ability to generate ideas to solve social and practical problems” (Homer-Dixon 1999, 109). Ingenuity is operationalized as the percentage of the population holding tertiary degrees as university-level education seems to capture the essence of ingenuity better than other measures like literacy rates. I use data by Barro and Lee (2010) on educational attainment and use the average percentage of people over the age of twenty-five with tertiary degrees between 1990 and 2010. Setting qualitative anchors is difficult, as existing thresholds are often policy goals, focusing on young people. Considering the data, I decided to set the threshold for fully in the set of countries with high levels of tertiary education at 13 percent, capturing five countries (Singapore, Japan, Uzbekistan, South Korea, and the Netherlands) and the threshold for countries to be fully out at 1 percent, capturing the following countries: Eritrea, Djibouti, Rwanda, Lesotho, Yemen, Burundi, and Haiti. The crossover threshold is set at 7 percent between Jordan (6.6 percent) and Mongolia (7.9 percent).
Analysis and Results
QCA does not assume a symmetry in causation, so while the presence of certain conditions might explain a conflict outcome, the absence of these conditions does not necessarily explain the absence of conflict. Therefore, separate analyses are needed for the conflict and nonconflict outcome. QCA differentiates between necessary and sufficient combinations of conditions. For an fsQCA, a condition is considered necessary if it has a very high consistency of 0.9 or higher (Schneider and Wagemann 2012, 143). Two measures—consistency and coverage—are given for each solution formula. Consistency indicates the degree to which cases share conditions, conveying the relationship between subsets of conditions and outcome. The coverage gives the degree to which the solution formula explains the set-membership scores in the sample. Conjunctural causation also assumes conditions to be interrelated.
Conflict Outcome
The analysis of necessary conditions for the conflict outcome gives one almost necessary condition: the absence of education has a consistency score of .89. As the membership values of the condition (x values) exceed the outcome values (y values) for a necessary condition, cases occur below or close to the bisector. This is mostly the case as can be seen in Figure 1.
Low levels of tertiary education as a necessary condition for a conflict outcome.
This means that there are no countries displaying a conflict outcome, that have high levels of tertiary education with the exception of Uzbekistan in the upper-left corner. This result supports the ingenuity hypothesis, which stated that countries experiencing conflict exhibited low levels of ingenuity.
In a second step, sufficient conditions are analyzed. The intermediate solution 2 yields two pathways to conflict (see Table 2). Both contain low levels of tertiary education, high levels of dependence on agriculture, and high poverty levels. 3 The first pathway combines these with high levels of corruption, while the second contains low quality of political institutions.
Intermediate Solution for Conflict Outcome.
Note: aUppercase denotes the presence of a condition, while lowercase denotes the absence of a condition. × denotes an AND combination of conditions, where both conditions must apply at the same time, that is, a case covered by the combination edu × DEP must have both, low levels of education and high levels of dependence on agriculture. + denotes an OR combination of conditions, where either one of the conditions to apply; that is, a case covered by the formula edu + DEP must either have low levels of education or high levels of dependence on agriculture.
The combination of high poverty levels and high levels of dependence on agriculture supports the economic situation of households hypothesis, which stated that countries with high levels of dependence on agriculture or high levels of poverty experienced conflict. The low level of tertiary education supports the human ingenuity hypothesis. As both pathways have these three conditions in common, they cover a lot of the same countries. Yemen, Pakistan, and Nepal are covered only by the first pathway including corruption, while Djibouti is covered only by the second pathway including low quality of political institutions.
With the exception of Pakistan and Kenya, all of the countries covered by this solution are among the United Nation’s list of least developed countries. The economies of these countries are characterized by a large agrarian sector and “a vicious circle of low productivity and low investment” (UN-OHRLLS 2012). Lacking human and institutional capacities and with low and unequally distributed incomes these countries incomes are often affected by bad governance and political instability. As a result, they are stuck in a “poverty trap” (UN-OHRLLS 2012), where a lack of human capacities and low productivity and a lack of opportunities for investment make it difficult to diversify the economy and create opportunities for income generation besides subsistence agriculture. Having a large percentage of the population dependent on agriculture is problematic in resource-scarce countries, where natural conditions impede the production of enough food without the use of advanced technology. The insufficient institutional capacities (represented by the high level of corruption in the first pathway and the low level of institutional quality in the second pathway) make it difficult for the state to mediate conflicts and deliver services to alleviate the situation. This shows how the various conditions act together and create a situation where poverty and a lack of opportunities to generate income increase incentives to join a rebel group and where there is little capacity in terms of human ingenuity to overcome scarcities and secure livelihoods as well as little institutional capacities to mediate conflicts, suppress violence or provide services and infrastructure needed to generate incomes.
The combination of high dependence on agriculture, poverty, and low levels of education raises questions regarding the role of economic development. However, replacing those three conditions with economic development shows that economic development is a necessary condition but not sufficient on its own. The intermediate solution includes three pathways 4 that are quite different from the intermediate solution presented previously. In addition, the solution including economic development has a lower consistency (.78 instead of .89), showing that the conditions in the analysis presented here do not merely act as a proxies for economic development.
Nonconflict Outcome
The analysis of necessary conditions found low levels of dependence on agriculture as the only condition that comes close to being a necessary condition (consistency of .89). High economic development has a consistency of 0.58, showing that dependence on agriculture gives a much better insight into which countries remain peaceful then the reliance on economic development in general. As Figure 2 shows, most cases have a lower membership in the nonconflict outcome than in the condition “low dependence on agriculture.” The most notable exception to this rule is Bhutan in the upper-left corner of the graph, which combines a peaceful outcome with high dependence on agriculture. It should be noted, however, that while Bhutan did not reach the twenty-five battle death threshold required to categorized as a conflict country in this study, it experienced significant ethnic strife in the early nineties (Giri 2004).
Low levels of agricultural dependence as necessary condition for a nonconflict outcome.
Analyzing the sufficient conditions for a nonconflict outcome gives four pathways: low levels of corruption combined with high levels of tertiary education (cor × EDU), low levels of corruption combined with low levels of agricultural dependence and low levels of poverty (cor × dep × pov), high quality of political institutions and high levels of tertiary education and low levels of agricultural dependence (PIN × dep × EDU) and low levels of ethnic exclusion, high quality of political institutions, low levels of corruption and low dependence on agriculture (eex × PIN × cor × dep) (see Table 3).
Intermediate Solution for Nonconflict Outcome.
In the various pathways that lead to a nonconflict outcome, some countries are covered by all four pathways: the Netherlands, Japan, Switzerland, Chile, and South Korea. However, in addition to these liberal democracies, the countries covered include autocratic countries such as Qatar, United Arab Emirates, Bhutan, and Kuwait, and countries with a low per capita income like Mongolia, Armenia, and Cape Verde. In these four pathways, different combinations of conditions work together to create the same result, which explains the different types of countries covered by the solution formula.
Discussion
My analysis sheds light on two factors that do not play a prominent role in the literature on resource scarcity and armed conflict: absence of agricultural dependence as necessary condition for a nonconflict outcome and absence of tertiary education as a necessary condition for conflict outcome. While lack of economic development has often been discussed as a root cause for conflict, the importance of the dependence on agriculture and how this affects resource-scarce countries in particular has been analyzed far less. This analysis shows that agricultural dependence combined with poverty and low levels of tertiary education leads to conflict and that the absence of dependence on agriculture is a necessary condition for a nonconflict outcome. While agricultural dependence is mentioned as an important factor in a number of qualitative studies (e.g., Farah, Hussein, and Lind 2002; Oketch and Polzer 2002), it plays a lesser role in quantitative studies. Some studies assume agriculture and poverty to be the link between environmental change or scarcity and conflict (e.g., Buhaug, Gleditsch, and Theisen 2008), but this is not often tested empirically. Zhang et al. (2007, 19217) explicitly hypothesize agricultural production as cooler periods are linked to a fall in agricultural production, which then instigated conflicts. Hendrix and Glaser (2007) argue that the negative effects of interannual variability in rainfall on conflict in Africa may be mitigated if agriculturalists were less dependent on rain-fed agriculture. In contrast to those findings, de Soysa (2002) found that having a higher percentage of arable land in a country decreased the risk of conflict. Finally, Urdal (2005) finds mixed results: while scarcity of potential cropland may have a pacifying effect, the combination of land scarcity and high rates of population growth increases the likelihood of internal conflict. Putting a focus on the role of subsistence agriculture research on conflict in environmentally vulnerable areas might help to give a clearer picture.
My results show an important role for tertiary education: low levels of tertiary education are a necessary condition for conflict, in particular in combination with dependence on agriculture and poverty. High levels of tertiary education are also part of two pathways to nonconflict. This supports the general mitigating effect of education that has been emerging in the wider conflict literature (Dixon 2009, 715-16). However, tertiary education has not received a lot of focus in this debate, which centers on indicators such as literacy rates or primary or secondary enrolment (Thyne 2006) or the secondary enrolment of males, focusing on opportunity costs of joining rebel groups and foregoing an education (Collier and Hoeffler 2004).
The notion of ingenuity plays an important role in the current debate on social consequences of climate change and its focus on adaptation and mitigation. Ingenuity is a very broad notion including aspects of political and legal institutions, education, technological innovation, and more. While the argument that societies need more ingenuity to solve the issues related to environmental change is certainly true, it is also rather unsatisfactory as it is difficult to pinpoint what ingenuity contains.
These results explain the contradictory results as the qualitative case studies generally focus on developing countries with high levels of subsistence agriculture and low levels of tertiary education, whereas quantitative studies generally do not restrict the sample in this way. In fact, with the exception of the study by Raleigh and Urdal (2007), all studies discussed previously that find a positive relationship between environmental change or scarcity and conflict are focused on Africa exclusively (Hendrix and Glaser 2007; Hendrix and Salehyan 2012; Raleigh and Kniveton 2012).
My results show under which conditions conflict occurs in resource-scarce countries, which does not give any insights into the bigger question of whether resource scarcity causes conflict. However, considering the conditions in the solution formula, they link up closely with the debate on climate change and conflict. There are three factors in vulnerability to climate change: exposure to climate change, sensitivity to climate change, and adaptive capacity (Scheffran et al. 2012, 870). Dependence on agriculture directly influences the sensitivity to climate change as climate and environmental change affect people more strongly where they depend on the land for their livelihoods. Tertiary education is directly linked to adaptive capacities, which could compensate for the impacts of climate change. So, while this study neither asks nor answers the question whether environmental change and resource scarcity cause conflict, the fact that the conditions it shows to be relevant are so closely related to environmental change suggests that there is a role for the environment in explaining conflict.
While introducing fsQCA as methodology with a specific understanding of causation to this field of research has therefore proved fruitful as it gives insight into the specific conditions that mediate the relationship between resource scarcity and conflicts, there are limits to the usefulness of fsQCA as a methodology. The fact that the number of possible combination of conditions increases exponentially with each new condition added limits the number of conditions that can be included in an analysis to four to seven four an intermediate n (Berg-Schlosser and De Meur 2008, 28). As a result, there is a risk of overlooking or not being able to include relevant conditions. In this case, a number of additional conditions might also yield interesting insights, for instance, unpacking the role of political institutions and focusing on aspects such as conflict-resolution mechanisms. Similarly, inequality in society and the access to and distribution of resources are interesting aspects to be included in future research. This might also explain those cases that are not covered by the solution formulas offered so far. In contrast to regression analysis, QCA does not yield any results to the importance or size of impact of certain conditions. So while tertiary education and agricultural dependence stand out from the conditions in the analysis as they are necessary as well as part of a sufficient condition, it is impossible to estimate their impact vis-à-vis other conditions.
However, the epistemology of conjunctural causation also brings new insights. This analysis includes examples of the benefits of including multiple pathways for the same outcome allowing for conjunctural and asymmetric causality. As scholars of resource scarcity have often argued that resource scarcity is not directly linked to conflict, but indirectly via a number of mechanisms, the assumption of multiple pathways is more appropriate for examining these links than the additive causality at the core of statistical analysis. The analysis of the nonoutcome, for instance, shows four quite different pathways to a nonconflict outcome, covering rather different types of countries. This shows that the same outcome (nonconflict) can be brought about by different combinations of conditions. Conjunctural causation allows QCA to explore the ways in which different conditions work together, such as the way in which conditions work together in least developed countries to create a “poverty trap” with a high risk for violence. While statistical analysis can include interaction effects, the number of these that can be included is very limited. The combination of high dependence on agriculture, high poverty levels, and low levels of tertiary education that is central to the solution of the conflict outcome would be difficult to include as an interaction term. Causal asymmetry allows for a situation where causes of the nonoutcome are not mirror images of the causes of the outcome as is the case in this analysis. This can provide new insights as asymmetric causalities cannot be detected by statistical analysis and case studies are more often focused on explaining an outcome rather than a nonoutcome. While closer in epistemology, QCA adds to the results of case studies by allowing for a systematic comparison across cases. While many case studies focus on analyzing the causes and mechanisms leading to conflict, QCA includes cases with a nonoutcome, which increases the generalizability of results. Overall, this analysis shows that QCA as a method can add new insights to the study of conflict. Using a specific notion of causal complexity, the focus of QCA research is often on questions that are slightly different to those answered by either statistical research or qualitative case studies.
Concluding Remarks
The introduction of configurational comparative methods such as fsQCA has allowed researchers new insights into fields where theoretical arguments employ a logic of causal complexity. One such issue is the link between resource scarcity and armed conflict, where empirical results have been inconclusive so far. I have argued that taking the political, social, and economic conditions into account can overcome this contradiction by employing a fsQCA to explain these contradictory arguments. After identifying thirty-one resource scarce cases (fifteen of which experienced conflict between 1990 and 2010), I have tested three hypotheses: the role of weak states, the economic situation of households, and the role of human ingenuity and found most support for the second and third hypothesis. This analysis shows that the assumption of causal complexity in QCA as method can add to the study of conflict. By analyzing the conflict and nonconflict outcome separately and allowing for different pathways to the same outcome, it adds a slightly different focus, which complements other methodological approaches.
Footnotes
Acknowledgement
I would like to thank the participants of Workshop 3 at the 2012 ECPR Joint Session and the participants at the VU political science staff seminar for their helpful comments on an earlier version of this article and two anonymous reviewers for their constructive feedback.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
