African range wars: Climate, conflict, and property rights

African Range Wars: Climate, Conflict, and Property Rights∗ Christopher K. Butler The University of New Mexico and Centre for the Study of Civil War, PRIO Scott Gates Centre for the Study of Civil War, PRIO and Norwegian University of Science and Technology (NTNU) June 22, 2010 Abstract This paper examines the effect of climate change on a type of armed conflict that pits pas- toralists (cattle herders) against each other (range wars). Such conflicts are typically fought over water rights and/or grazing rights to unfenced/unowned land. The state is rarely involved directly. The Rangeland of East Africa is a region particularly vulnerable to drought and livestock diseases associated with climate change. To analyze the possible effects of climate change on pastoral conflict, we focus our analysis on changes in resource availability con- trasting cases of abundance and scarcity. The role of resources is further contextualized by competing notions of property rights, and the role of the state in defining property and associ- ated rights. Using the analytical framework of a Contest Success Function (CSF), we formally demonstrate that conflict is generally increasing with resource abundance contingent on prop- erty rights bias. ∗ Paper prepared for presentation at “Climate Change and Security”, 250th Anniversary Conference organized for the Royal Norwegian Society of Sciences and Letters (Den Kongelige Norske Videnskabers Selskab, DKNVS) Trondheim, Norway, 21-24 June 2010. This project has in part been funded by the “Poverty and Peace Program” of the Research Council of Norway and the EU 6th Framework project, “Polarization and Conflict”. 1 1 Introduction Dealing with impending change has long brought out the alarmists in us. Indirectly relevant to the climate change debate is Thomas Mal thus’ argument that a geometrically expanding population would outstrip arithmetically expanding agricultural production, leading to starvation, conflict, and war (1826). Such alarmist arguments aren’t unique to the nineteenth century. With respect to climate change and conflict, James Lee (2009) argues that inter-state conflict (though perhaps short of war) will be more prevalent at higher latitudes over access over newly exposed resources while intra-state will be more prevalent near the equator over scarcer resources.1 We try to step back from these alarmist arguments by presenting a model that points to situations where we should expect more conflict as well as situations where we should expect less conflict. The idea that resources affect conflict is not unique to the climate change and conflict literature. Resources are also linked to conflict in a literature of its own. On the one hand, the “resource curse” argument is that countries endowed with plentiful resources are cursed to have higher levels of conflict than countries that are not so plentifully endowed (Auty 1993; de Soysa 2002; Collier and Hoeffler 2003). On the other hand, “environmental scarcity” scholars contend that conflict is due to the strains of competition by many people over few resources (Baechler 1998; Homer-Dixon 1999; Ohlsson 1999; Kahl 2006). These arguments have mixed results in the empirical literature. Few if any large-N empirical studies have been able to establish any relationship between scarcity and inter-state war or intra-state war or conflict (Raleigh and Urdal 2007; Hendrix and Glaser 2007; Nordås and Gleditsch 2007; Buhaug, et al. 2009; Theisen, et al. 2010). In contrast, there does appear to be some case study support for the scarcity argument for inter-communal violence (Mkutu 2008; ), but these cases are strongly contradicted by empirically based case studies such as (Witsenburg and Adano 2007, 2010; Theisen 2010). If there is going to be a relationship between ecological degradation (leading to scarcity) and armed conflict, it should be most strongly evident in the form of intercommunal or ethnic violence. More particularly it should be most evident among those populations most directly affected by such scarcity, farmers and herders. The pastoralists of the range lands of East Africa are among the most vulnerable to drought and stock disease (e.g. rinderpest which can devastate a herd) and other disasters associated with climate change. Pastoral conflict, mostly in the form cattle rustling as the main form of violence, thus constitutes a critical test of the environmental scarcity proposition to test the analytical logic underlying the debate. In this paper we focus on the mechanisms by which resource scarcity would lead to armed conflict between pastoral groups. While our analytical approach is a formal analytical model that could be classified as political economy, our approach is similar to the “political ecology” approach (Bassett 1988; Le Billon 2001; Turner 2004; Benjaminsen and Ba 2009; Benjaminsen et al 2009). Advocates of the “political ecology” approach stress that conflicts between agrarian groups should be understood as more than just resource conflicts that the politics of land plays a critical role. We reflect this perspective by focusing on the biased but limited role the state plays in establishing, protecting and recognizing property rights in Africa’s rangelands. More particularly we analyti- 1 In fairness, Lee does say that conflict due to climate change is not inevitable, but will depend on how we react to these changes. 1 cally examine the role of resource scarcity and abundance as affected by property rights protection and bias in this protection in shaping pastoral violence. Through our analytical framework we demonstrate that resources alone (whether scarce or abundant) only go so far in explaining this form of communal violence. We show how differences in property rights protection and the state’s bias in protection of these rights can explain why in most cases pastoral violence is associated with resource abundance, but also the conditions when resource scarcity leads to violence. 1.1 Resource Allocation and Conflict We analyze conflict through the development of a contest success function model. Contest suc- cess functions (CSFs) offer a flexible way to model the dynamics of conflict. CSFs are economic models of conflict. Rather than featuring only the technologies of production as most other eco- nomic models do, they also incorporate the technologies of appropriation and war. Contest success functions allow us to model both aspects of an economy, the productive and the predatory. In an environment of a weak or fragile state, contracts are weakly enforced. Property rights, in particu- lar, tend to be weakly enforced at best. The institutions that serve to regulate the economy work inefficiently and perform even more poorly when violence is endemic. Consequently, conventional models of economic production fail to accurately model the political-economy of violent conflict. Indeed, the original idea of a CSF is that actors choose how they are going to allocate their in- dividual resources in the total absence of property rights. Resources can be allocated either into production or appropriation. The allocation of this wealth is what turns money into guns (for ap- propriation) or butter (production). In this respect, the CSF modeling approach offers an effective way to model pastoralist-sedentary agriculturalists conflict. Conventional production models assume perfect property rights and enforceability of contracts. Hirshleifer purposely examines the “dark side” of economic activity in which the opposite is true on both counts. The two types of model are at opposite ends of an institutional spectrum. Part of our purpose here is to examine the full range of institutional possibilities. Rather than the stark contrast of binding, enforceable contracts versus the complete absence of moderating institutions, we examine interactions of non-state actors under the imperfect gaze of the state. State institutions are neither perfect nor absent. Instead, we examine how the biases inherent in state institutions and their relative strength can be significant factors accounting for the level of conflict between non-state actors. One point to note, we do not model cases in which survivability of a group is at stake. For example, the choice between starvation and fighting has been modeled by Grossman and Mendoza (2003) and the role of war when survivability is at stake (Bas and Coe 2010). In the real world, group survivability is rarely at stake (even in the context of fierce conflict). We presume groups survive conflict. This is hardly a heroic assumption given that we are modeling a very low scale form of conflict, cattle raiding. 1.2 Property Rights and (Not Necessarily Ethnic) Conflict While the traditional economic production model assumes perfect property rights, the basic CSF model assumes the complete absence of property rights. In a companion paper (Butler and Gates 2 2010), we build on a particular CSF model (Neary 1997) so we can examine intermediate and potentially biased property-rights regimes. Of particular interest to us is that different groups might be differentially protected or favored by the government.2 We focus in particular on one type of conflict—communal violence between pastoralists (herding groups). In particular, we examine the level of property rights protection (PRP) and the government’s bias in property rights enforcement (Bias). In modeling PRP as a variable, we are able to examine the interactions of non-state actors in a weak state environment in which state institutions are neither perfect nor absent. Our CSF model demonstrates that increasing property rights protection reduces the effective- ness of fighting, which implies increasing the equilibrium allocation of productive effort. In other words, as PRP is increased, groups accordingly allocate more effort to butter rather than to guns. Our model also accounts the potential bias towards one interpretation of property rights over an- other interpretation. When two different groups are defined by their different uses and definitions of property, one group is inherently favored over another. Such bias plays a regular role in farmer- herder conflicts as the two groups by definition regard property in fundamentally different ways. Bias and property rights protection interact to produce a non-monotonic result affecting the level of conflict in a society. More particularly, if a society has a moderate level of PRP, but some de- gree of bias away from equity, an increase in PRP can result in either a decrease or an increase in the amount of fighting between the two groups. Thus, simply increasing PRP without addressing equity and bias issues can actually increase the risk of armed conflict between pastoralists and sedentary agriculturalists. By bias we mean a pattern of interpretation of property rights that favor one group over another. For example during the Colonial period, white settlers in East Africa were systematically advan- taged at the expense of the indigenous populations. Such biases, however, did not end with inde- pendence. Post-colonial governments have by and large maintained a clear bias towards sedentary farmers over pastoralists. Property laws in particular granted title to land rather than the traditional notion of entitlement to use the land but not to own it per se. Pastoralists are particularly dis- advantaged by this conceptualization of property rights, which strongly favors sedentary farmers with set plots of land on which they grow crops as opposed to the open rangeland grazing of pas- toralists. Property rights bias can also occur between pastoralist groups when territory is divided between groups granting exclusive rights to one particular group, excluding others from grazing rights. Enforcement of property rights need not be biased, but it usually is. Bias in property rights protection relates closely with the notion of horizontal inequality (Stew- art 2000; Murshed and Gates 2005; Østby 2008). The notion of horizontal inequalities between groups, classified by ethnicity, religion, linguistic differences, tribal affiliations, caste, etc., is dis- tinguished from vertical inequality, which is the inequality between individuals within an other- wise homogeneous population. Horizontal inequality usually stems from historical discrimination or systematic patterns of economic exclusion of a particular group, typically distinguished by eth- nicity or religion identities. What really matters are the differences between the two groups, not the within group inequalities in each of these individual identities (Murshed 2009). Bias in property rights protection constitutes one aspect of economic horizontal inequality. 2 The model merely presumes the prior existence of two groups. These could be Sneetches with or without stars on their bellies as far as the model is concerned (Suess 1961). 3  Our paper will examine how changes in the resources available to the groups might change their equilibrium splits between production and appropriation. Before examining that core question, we now discuss how climate change may affect resource levels. 1.3 Climate Change, Resources, and Conflict Strictly speaking this paper is about weather changes, particularly about droughts, which in turn affects resource availability, and the effect on pastoral conflict. To study the phenomenon that we are investigating, we need not look at climate change per se. According to the World Me- teorological Organization, climate is the statistical description (mean and variability) of surface weather conditions such as temperature, participation, and wind over a period of 30 years. “Cli- mate change is a statistically significant and persistent (decades or longer) variation in the mean state or variability of climate (IPCC 2001)” (Cioffi-Revilla, et al 2010). Climate change will affect the weather, thereby increasing the frequency of droughts. But we need not have climate change to have drought. Nevertheless, with climate change, especially a hotter and drier East Africa, will lead to more frequent drought. By analyzing the relationship between droughts and pastoral conflict, we can shed some light on the relationship between climate change and conflict. What constitutes resources that can be allocated is an important question for our analysis. Some of these resources are affected by changes in weather while others are not. Within limits discussed below, livestock will be more plentiful when fresh water is more available. Livestock can be bred, or can be sold and the proceeds reinvested. Cattle can also be substituted for goats in times of drought or traded for cattle in wetter times (Anderson 2002). A group can devote its time and effort to herding and breeding or they can rustle cattle from other groups. They can produce or they can appropriate. We are making an explicit assumption that more rain leads to more resources for allocation and that less rain leads to fewer resources. However, two extremes need to be discussed. On the one hand, an over-abundance of water (whether through heavy rain or through excessive glacial runoff) cannot be effectively harnessed without significant public infrastructure and can, in fact, reduce productivity of the land by destroying crops and killing livestock. Thus, above some limit, more rain does not lead to more resources but is likely to lead to less resources. Such flooding is also associated with such diseases as Rift Valley Fever, foot rot and diseases born by biting flies. El Nio floods in 1997-98 were disastrous for pastoral communities in northern Kenya and southern Somalia (Little, et al 2001). On the other hand, a short-term shortage of water is different from a long-term drought. It is conceivable for a community to live off of previous reserves given one dry season. This would fit our “less rain, fewer resources for allocation” assumption. There is a minimum subsistence limit regarding this. We presume that the two groups are in sustained contact with each other. If minimum subsistence isn’t being achieved, however, among a variety of coping mechanisms a community may decide to migrate away from their traditional grazing area. When they come in contact with some other community, their interactions would be guided by the fact that they have little to invest but their time in an appropriative way. Thus, we would expect mass migrations to lead to increased conflict as those driven off their traditional grazing lands are forced to forage on land claimed by other groups. 4  Several studies have examined the relationship between rainfall (or more particularly drought) and pastoral conflict in East Africa. Meier, Bond, and Bond (2007) find no relationship between rainfall deficiency and violence in their study of the borderlands of Uganda, Kenya and Ethiopia. They further show in their examination of seasonal variation over a two year period that a greater abundance of vegetation is associated with an increase in the frequency raids. Witsenburg and Adano (2010) analyze monthly rainfall and cattle raiding data from 1960–2006 in the Marsabit district of northern Kenya and find that wetter years on average are associated with more than twice as many killed as compared to drier years (50 vs. 23). They argue that livestock raiding, especially violent episodes, result from opportunistic behavior. Attacks occur during wet seasons when water is abundant, pastures are lush, and livestock (particularly cattle) are abundant and in good health. When rangeland resources are abundant, labor will tend to be in surplus, which provides a large pool of young men to recruit into a cattle raiding party. Moreover, abundant vegetation makes it easier to hide and take advantage of the element of surprises. Readily available water and vegetation also make it easier to drive the stolen cattle long distances. And given that the cattle are healthier, they are better able to make a long trek. Rainfall also serves to wash away the tracks. In contrast, in times of drought, herds tend to decrease in size and the herds shift from cattle to camels and goats which are better adapted to dry conditions. Herding groups will be focused on survival and will be moving with their herds over wide areas. Labor demand in dry times is taken up looking for far-away pastures instead of engaging in cattle raids (Witsenburg and Adano 2010). Taking animals greater distances to pasture and into areas with such unusual attendant dangers placed great demands upon herd management, particularly in terms of labor requirements. Every male in the family might accompany the stock on these move- ments, sometimes remaining away from the permanent homestead for two or three months. Close friends, age-mates, or clansmen combined their herds for these trips, for safety and companionship. Extra labor was also demanded to find food for cattle when grazing was not available (Anderson 2002: 82). Moreover, stealing cattle is less lucrative during drought. In an analysis of the relative prices of commodities in pastoralist areas of the Horn of Africa Mkutu (2000: 17) reports that the price of cattle drops in drought years, whereas the prices of grains necessary for survival rise. Even if you don’t steal cattle during a drought, why fight over a dry well? Theisen (2010) finds preliminary confirmation in a statistical analysis of geo-referenced data that violent clashes between groups in Kenya do not occur in years when water is scarce, but when it is plentiful. These studies for the most part do not account for the role of property rights in their analysis. To be fair many of these studies, such as Theisen (2010) do look at other variables (e.g. elec- tions) that may spark violence between groups, but property rights are generally not considered in the analysis of pastoral conflict. Political economists such as North (1990), Knack and Keefer (1995; 2003) have persuasively shown how property rights protection can affect economic growth and development. But these works ignore the role of bias in protection. Neutral application and interpretation of property rights is often impossible. One group invariably benefits at the expense of another. Indeed the establishment of entitlement or exclusive use of land, will most often be biased. This bias is critically important for understanding violent communal conflict. 5  Anderson (2002) examining the colonial history of Baringo in the Rift Valley of Kenya from 1890 to 1963, describes the violence between the white settlers and the pastoralists that was par- ticularly prevalent in the 1920s. Baringo’s range war was a desperate struggle for survival, revolving around the con- tested meaning of land ownership and access to resources. For the European settler, it was simply a matter of upholding the law. Facing financial ruin because of the reces- sion and drought, his very future depended upon the protection of his property from encroachment. For the African herder, this was irrelevant. His choice was equally stark: either he remained in the reserve and watched his cattle die or he crossed the boundary to take advantage of better pasture. Even when confronted by the threat of arrest and fining and imprisonment, the majority of herders were willing to take the risk (Anderson 2002: 102). Since colonial times, the pastoral policy discourse in Kenya and Tanzania as well as other East African countries has been shaped significantly by a modernist ideology (Raikes 1981). Modern husbandry it is said entails land titling, enclosing the commons, and establishing permanent set- tlements Benjaminsen et al 2009). Such policies fundamentally alter land use policy by affecting the nature of property rights. In Kenya, trust land reserved for pastoralists has been controlled by county councils, but governed by customary law. Access to water in particular is governed by traditional use rights. “Water sources are usually owned by individuals or clans, but use rights traditionally override ownership rights. This means every nomad has the right to use a well, also on the land governed by a neighboring ethnic group” (Witsenburg and Adano 2009: 532). During drought, access to water was traditionally successfully negotiated. “With the creation of bound- aries, whereby ... new divisions and locations are meant to mark ethnically ‘pure’ territories, it is increasingly hard to maintain traditional arrangements of water and land sharing. The violence be- tween the Gabra and Borana in Turbi, between Redille and Gabra around Medatte (or Mayidahad), and between the Gabra and Turkana in Moite ... can be seen in that light. In all these places, water points were ’territorialized’ after the creation of new boundaries, and traditional use rights were violated (Witsenburg and Adano 2009: 532). Furthermore, as reported in Witsenburg and Adano (2009) from an interview taken during their fieldwork: Raiding today is not for the same reason as in the past. Today, it is a way to express resentment against the authorities, for instance in the creation of administrative bound- aries. For the nomads there is no other way to express their resentment. You cannot, like in Europe, demonstrate in the street and create media coverage. People here are out of communication. They want to show their resentment, through raiding neighboring groups whose polit- ical leaders are powerful, and who are seen as the oppressors. If you go out fighting over land, and you see goats, there is no way you will leave them (Hilary Halkano interviewed by Witsenburg and Adano (2009: 531-32)). The process of exclusion serves as the motivation to engage in violence (Cederman, et al 2010), but subject to resource availability. Given the positive relationship between resources and fighting, 6 we should expect to see a wealthy excluded group should fight more than a wealthy included group, which should fight more than a poor excluded group. We demonstrate this formally in the next section. 2 Modeling Resource Fluctuations, Property Rights, and Con- flict We answer our question theoretically through the use of a contest success function (CSF). CSFs have been used by others to model many aspects of conflict. Our model has several assumptions common with other CSFs. Each group optimizes its welfare given the anticipated behavior of the other group. Each group has some initial resources (Ri > 0) that are then allocated into productive and appropriative effort. The appropriative effort is conceptualized as fighting effort, Fi ∈ [0, Ri ]. In the absence of enforceable contracts, all productive effort is assumed to create a collective income (I) that is divided between the two groups according to their appropriative effort. Collective income is assumed (for simplicity) to be the sum of productive effort. I = (Ri − Fi ) + (Rj − Fj ) (1) To answer our question, we adopt a modified version of a CSF model first presented by Neary (1997). The proportion of collective income that group i gets (pi ) is a function of its fighting effort divided by the total fighting effort of both groups. The η terms (assumed to be strictly positive and non-zero) allow for definition of this function when fighting effort is zero.3 We conceptualize of η as the enforcement of property rights. It is through these additional terms that we will analyze property rights and bias later in the paper. Fi + ηi pi = (2) (Fi + ηi ) + (Fj + ηj ) Given this division mechanism, each group maximizes its share of the collective income (Ii = pi I). Fi + ηi Ii = [(Ri − Fi ) + (Rj − Fj )] (3) (Fi + ηi ) + (Fj + ηj ) When ηi = ηj and neither actor invests in fighting effort, collective income is divided equally (i.e., pi = 0.5). We argue more generally that in situations of no conflict, the division of income would be premised on an expected division of collective income based on boundaries or norms. In this regard, the ratio ηi /(ηi + ηj ) can be interpreted as the no-conflict division to group i when Fi = Fj = 0. This no-conflict division is offset in practice by investments in fighting effort. 3 This alters Neary’s (1997) CSF by subscripting the η terms. Also, there is usually a “military decisiveness” parameter included in such CSFs as an exponent on each fighting effort parameter. As the effects of this parameter are not a focal concern and add considerable complexity to the mathematics, we essentially assume that this parameter is unity throughout. 7  Both groups are assumed to make their allocation decisions simultaneously. Group i’s reaction curve is found by taking the derivative of (3) with respect to Fi and then solving for Fi . q Fi = (Fj + ηj )(Ri + Rj + ηi + ηj ) − Fj − ηi − ηj (4) Group j’s reaction curve is found by the same process. Substituting group j’s reaction curve into (4) and solving for Fi again, we find group i’s equilibrium level of fighting effort as a function of initial resources and property rights enforcement, the η terms. This holds so long as the joint equilibrium is an interior solution (in which both groups invest in some fighting effort, but less than their resource allocation). If Fi∗ and Fj∗ is jointly an interior solution (i.e., Fi∗ ∈ (0, Ri ) and Fj∗ ∈ (0, Rj )), then (Ri + Rj − 3ηi + ηj ) Fi∗ = (5) 4 (Ri + Rj − 3ηj + ηi ) Fj∗ = (6) 4 If we do not have an interior solution, then we must consider what Fi∗ looks like when Fj is at one extreme or the other. When Fj∗ = 0, group i’s equilibrium fighting effort is found by substituting that value into (4). This is given by (7). q q Fi∗ = (ηj ) (Ri + Rj + ηi + ηj ) − ηi − ηj (7) At the other extreme, one group may have fighting effort equal to its initial endowment. This can occur when its resources are low and its level of protection is also low. When group j has this extreme level of equilibrium fighting effort, group i’s equilibrium fighting effort is found by substituting Fj = Rj into (4). This is given by (8). q q Fi∗ = (Ri + Rj + ηi + ηj ) (Rj + ηj ) − Rj − ηi − ηj (8) 2.1 Property-Rights Regimes Given the large number of free-floating parameters, it is difficult to interpret these equilibria. While we are interested in how changes in resources affect fighting effort, we also want to understand how property rights affect these effects. That is, do similar changes in resources have the same effects in altering fighting effort under different property-rights regimes. In our companion paper, we discuss two property rights parameters. The level of property rights protection (PRP), ρ = ηi + ηj , is related to how much the government invests in property rights. Merely investing in property-rights protection, however, is not sufficient to defining the property- rights regime. We also need to know how equitable that protection is or, alternatively, how biased that protection is toward one group. Focusing on bias, we define βi = ηi /ρ as the bias in property- rights protection with respect to group i. Given just two groups, we can define equitable protection in terms of ηi = ηj . Thus, when βi = 0.5, property-rights protection is equitable. When βi < 0.5, 8 property-rights protection is biased against group i and, conversely, when βi > 0.5 property-rights protection is biased in favor of group i. The main result of our companion paper normalizes Ri + Rj = 1 and shows how total societal conflict (TSC, that is, total fighting effort) changes over different levels of property-rights protec- tion and bias. In particular, we find that equilibrium levels of TSC are lowest when high levels of PRP are combined with equitable PRP. Having high levels but biased PRP produces more TSC. Having equitable PRP but low levels of PRP also produces more TSC. Figure 1 provides a visual representation of this result. Figure 1: Total societal conflict by ρ and βi We were able to normalize Ri + Rj = 1 (rather than assuming Ri = Rj = 0.5) by assuming we were dealing with either an interior solution or an edge solution in which one group invests nothing fighting (Fi ∈ [0, Ri ) and Fj ∈ [0, Rj )). Within these constraints, one group’s resources decreasing while the other group’s resources increase accordingly produces the same level of TSC. Such a perfect pair of changes in resources is unlikely to occur in reality and was not the focus of that paper (which assumed group resources remained constant). A convenient way of building on that result and making sense of how TSC is effected by both changes in resources and changes in property-rights regime is to start with the assumption that Ri = Rj = 0.5 and then alter Ri while holding Rj = 0.5. Figure 2: Total societal conflict by Ri and βi given ρ = 0.5 and Rj = 0.5 Figure 2 holds Rj = 0.5 and ρ = 0.5.4 This leaves Ri and βi as free parameters. TSC is lowest in this figure when PRP is equitable and group i’s resources are near zero. When group i’s resources are relatively small, more biased PRP produces more TSC. If the bias is in favor of group i, group j contributes more toward TSC; if bias in against group i, group i contributes more toward TSC. For higher levels of group i’s resources, TSC increases linearly with total resources until an edge solution kicks in. At that point, Fj∗ = Rj and TSC goes up at a slower rate with further increases in Ri (without asymptote). This edge solution is reached quicker when PRP is biased in favor of group i. Once reached for a given level of Ri , TSC is higher when PRP is biased against group i. The main formal result here is that a relatively wealthy group under a property- rights regime that is biased against them will expend a greater amount of their resources in fighting effort. 4 These parameters wore chosen arbitrarily but the general picture holds for other parameter choices. If Rj is higher, TSC is higher. If PRP is higher, TSC is lower. 9 3 Discussion This paper examined the effect of climate change on pastoralist conflict (range wars). The Range- land of East Africa constitute a region particularly vulnerable to drought, which is associated with climate change. To analyze the possible effects of climate change on pastoral conflict, we focus our analysis on changes in resource availability contrasting cases of abundance and scarcity. We adopt a “political ecology” approach and feature the role politics of land plays in pastoral conflicts. In particular we focus on property rights protection and the bias in interpretation and administra- tion of property rights protection. Violence between pastoral groups should be understood as more than just resource conflicts. Moreover, we demonstrate that any analysis of climate change and conflict needs to account for property rights protection and even more importantly the bias in these property rights. In this way we account for the limited role the state plays in establishing, pro- tecting and recognizing property rights in Africa’s rangelands. Using the analytical framework of a Contest Success Function (CSF), we formally demonstrate that conflict is generally increasing with resource abundance contingent on property rights bias. 10 4 References Benjaminsen, Tor A. 2006. “States, Scarcity, and Civil Strife in the Developing World”, Jour- nal of Peace Research 43(6): 754-754. Benjaminsen, Tor A., Faustin P. Maganga, and Jumanne Moshi Abdallah. 2009. “The Kilosa Killings: Political Ecology of a Farmer-Herder Conflict in Tanzania”, Development and Change 40(3): 423-445. Hirshleifer, Jack. 1995. “Anarchy and Its Breakdown.” Journal of Political Economy 103(1):26– 52. Hirshleifer, Jack. 2001. The Dark Side of the Force: Economic Foundations of Conflict Theory. Cambridge: Cambridge University Press. Homer-Dixon, Thomas, 1999. Environment, Scarcity and Violence. Princeton, NJ: Princeton University Press. Kahl, Colin, 2006. States, Scarcity, and Civil Strife in the Developing World. Princeton: Prince- ton University Press. Knack, Stephen, and Philip Keefer. 1995. “Institutions and Economic Performance: Cross- Country Tests Using Alternative Institutional Measures.” Economics and Politics 7(3): 207-227. Lee, James R. 2009. Climate Change and Armed Conflict: Hot and Cold Wars. Routledge Studies in Peace and Conflict Resolution Malthus, Thomas Robert. 1826. An Essay on the Principle of Population, 6th edition. London: John Murray. Accessed via http://www.econlib.org/library/Malthus/malPlongCover.html. Murshed, S. Mansoob and Scott Gates. 2005. “Spatial-Horizontal Inequality and the Maoist Insurgency in Nepal” Review of Development Economics 9(1): 121–134. Neary, Hugh N. 1997. “Equilibrium Structure in an Economic Model of Conflict.” Economic Inquiry 35(July): 480-94. Østby, Gudrun. 2008. “Polarization, Horizontal Inequalities and Civil Conflict?”, forthcoming in Journal of Peace Research 45(2): XXX-XXX. Seuss, Theodore. 1961 The Sneetches and Other Stories. New York: Random House. Stewart, Frances. 2000. “Crisis Prevention: Tackling Horizontal Inequalities.” Oxford Devel- opment Studies 28(3): 245-62. Witsenburg, Karen and Adano Wario Roba. 2007. “The Use and Management of Water Re- sources in Kenya’s Drylands: Is There a Link between Scarcity and Violent Conflicts?” in Conflicts over Land and Water in Africa, edited by Bill Derman, Rie Odgaard, and Espen Sjaastad. James Curry: Oxford. 11 Figure 1: Total societal conflict by ρ and βi Figure 2: Total societal conflict by Ri and βi given ρ = 0.5 and Rj = 0.5