Scholarly article on topic 'Political hierarchies and landscapes of conflict across Africa'

Political hierarchies and landscapes of conflict across Africa Academic research paper on "Social and economic geography"

Share paper
Academic journal
Political Geography
OECD Field of science
{Conflict / Africa / Scale / "Spatial analysis"}

Abstract of research paper on Social and economic geography, author of scientific article — Clionadh Raleigh

Abstract Almost all African states experience substantial and widespread political insecurity in a variety of forms. This analysis explains how relationships between groups and governments create incentives and disincentives for distinct forms of political violence to emerge. It argues that ethno-regional communities across Africa are situated within a power hierarchy that determines their relative importance to, and inclusion in, regimes. A dynamic power landscape emerges from relative group positions. Various positions within a hierarchy are associated with particular dominant forms of organized political violence as groups challenge political elites, but are bounded by their goals and characteristics. A failure to consider the political hierarchies and landscapes operating within African states has led to an under specification of the causal mechanisms driving different forms of violence, and an overstatement of benefits from declining civil war rates and inclusive governing coalitions.

Academic research paper on topic "Political hierarchies and landscapes of conflict across Africa"


Contents lists available at ScienceDirect

Political Geography

journal homepage:



Political hierarchies and landscapes of conflict across Africa

Clionadh Raleigh

University of Sussex, UK


Almost all African states experience substantial and widespread political insecurity in a variety of forms. This analysis explains how relationships between groups and governments create incentives and disincentives for distinct forms of political violence to emerge. It argues that ethno-regional communities across Africa are situated within a power hierarchy that determines their relative importance to, and inclusion in, regimes. A dynamic power landscape emerges from relative group positions. Various positions within a hierarchy are associated with particular dominant forms of organized political violence as groups challenge political elites, but are bounded by their goals and characteristics. A failure to consider the political hierarchies and landscapes operating within African states has led to an under specification of the causal mechanisms driving different forms of violence, and an overstatement of benefits from declining civil war rates and inclusive governing coalitions.

© 2014 Elsevier Ltd. All rights reserved.


Article history: Available online

Keywords: Conflict Africa Scale

Spatial analysis

Almost all African states experience substantial and widespread political insecurity in a variety of forms throughout their territory (Raleigh, Linke, Hegre, & Carlsen, 2010). This analysis highlights the variety in, and explanations for, political violence. It argues that ethno-regional communities across Africa are situated within a power hierarchy that determines their relative importance to, and inclusion in, regimes; these relative group positions create a dynamic power landscape. Positions within governance hierarchies are associated with a dominant form of conflict as relationships between governments and groups create incentives and disincentives for distinct forms of political violence. A failure to consider the political hierarchies and landscapes operating within African states has led to an under specification of the causal mechanisms driving different forms of violence, and an overstatement of benefits from declining civil war rates and inclusive governing coalitions.

Many conflict researchers explore explanations for conflict variation and occurrence. Studies distinguish whether states experience a 'revolutionary' or 'separatist' civil war (Buhaug, 2006); whether communal or 'livelihood' violence is triggered by environmental change (Raleigh, 2010a; Raleigh & Kniveton, 2012; Straus, 2011); or how 'warlord' violence, characterized by high rates of criminal activity and violence against civilians (see Bates, 2008; Reno, 1998), is the basis of the 'new war' thesis (Kaldor, 1999). The underlying motive often differentiates violence: the greed versus grievance literature proposes that 'sons of the soil' contests (Fearon, 2006) differ from those aiming to control the state

E-mail address: 0962-6298/© 2014 Elsevier Ltd. All rights reserved.

and access rents and resources (Collier & Hoeffler, 2002, 2004; Le Billon, 2001). Recent research suggests that a key source of heterogeneity in civil wars is rebellion technology, itself a function of the relative balance of power between opponents (Kalyvas & Balcells, 2010). Violence types differ on levels of public support, funding sources (Weinstein, 2007), or by the use of child soldiers and sexual violence (see Dixon, 2009). These studies further our understanding about how specific forms of violence are produced, but no explanation yet offers specific distinctions between types of violence, how those forms are deliberately shaped by groups, found in specific locations, and strategically scaled on the local, regional and national levels.1 Using present theories of conflict, how and why multiple forms of violence emerge in countries such as DR-Congo, Nigeria and Kenya is unexplained. This is a quite serious limitation.

This article argues that multiple, distinct forms of political violence co-occur within states but have limited rates of spatial overlap. Heterogeneous violence is a result of subnational political processes that triggers its emergence.

A broad definition of political violence allows for an examination into its heterogeneity, spatial characteristics and variation in occurrence. 'Political violence' is the use of force by a group with a political purpose or motivation, often designed to secure resources, and access or alter paths to power. A key determinant of African political violence is that political, ethnic, religious or regional groups and areas are targeted.2 For the purpose of this analysis, non-state political violence is classified into three forms: civil wars, militia violence, and communal conflict. These forms differ based on

group goals, impacts on national security, thresholds and scales of violence: Civil wars involve a primary cleavage of rebels against an established government, where the objective is to replace the regime or establish a separate state (Gleditsch, Wallensteen, Eriksson, Sollenberg, & Strand, 2002; Sambanis, 2004). These enduring contests often subsume large areas of a state and result in high military and civilian fatalities. Political militias operate as 'private armies' for political elites broadly including regimes (e.g. Janjaweed in Sudan), members of governments (e.g. Mungiki in Kenya); and political opponents (e.g. Boko Haram in Nigeria) (Raleigh, 2014). These armed groups are often used by politicians to compete over access to power; conflicts are frequently limited in spatial scope and fatalities are significantly fewer than civil wars. Communal contests are local encounters between identity-based groups including ethnic, regional, religious or livelihood communities. These conflicts emerge over territorial disputes, local power disparities, resource access and historical disagreements. These three categories represent a continuum of organized political violence from local uprisings to coordinated violence over state


Each form displays differences in rates of activity, targeting, and spatial patterns: according to the Armed Conflict Location and Event Data (ACLED) (Raleigh et al. 2010) from 1997 to 2013, rebel groups in civil wars spend half of their active time engaging with military forces, and one quarter on civilian attacks; the activity of individual groups last for over a year. In contrast, militias spend the majority of active time attacking civilians, but these acts result in fewer fatalities per event, and discrete groups endure for less than half a year with no significant change in governments resulting from their activity. Communal conflict is often repeated and seasonal; it is divided between battles with ethnically or regionally identified opponents, or attacks on civilians within those communities. High fatality rates are common across communal contests. Across types, individual groups may practice multiple modalities and tactics of violence, including terrorism (threats and randomized violence, often upon civilians), electoral violence, rebellion, and insurgency strategies.

These characteristics not only suggest that types of political violence are different, but that they are potentially non-substitutable, as each form is directed towards achieving an exclusive political goal. Furthermore, the occurrence of each type is spatially distinct, and the level of overlap between types varies within the same state (see Table 1)4: overlap is low between militia and communal violence, lowest between civil war and communal violence, and highest between civil war areas and militia activity.

Distinct conflict types are found in separate locations, but they co-occur within states. Particular constellations of temporal cooccurrence are common: in weak states such as DR-Congo, Somalia or Nigeria, civil wars, political militia activity and communal

Table 1

Overlap between active locations for three forms of political violence.

Spatial overlap Civil war Political militia Communal militia

Annual Civil war 100% 17% 6%

Political militia 20% 100% 9%

Communal 17% 22% 100%


Static Civil war 100% 31% 20%

zones (15 km)a

Political militia 35% 100% 30%

Communal 36% 47% 100%


a Static statistics are calculated by recording if any form of violence occurred and overlapped within a grid square at anytime between 1997 and 2011. A15 km buffer is placed upon all conflict points to account.

violence are found in different areas within the same state. In Ivory Coast and Chad, civil wars and militia violence simultaneously occur throughout the state. In multiple countries experiencing unstable democratic transitions (e.g. Kenya, Tanzania), political militia activity and communal violence occur with limited local overlap. The subnational variation in occurrence suggests that the form of conflict in which communities engage may be in response to local characteristics, and the presence of a heterogeneous mechanism creating incentives and disincentives for particular forms of violence to emerge.

This article investigates how subnational regime practices structure the motivations and mechanisms of ethno-regional communities to engage in different forms of co-occurring political violence within and across African states. Much of the intra-state conflict literature uses 'governance' as a blunt instrument, explaining local differences in violence as a result of state capacity vacuums (Hendrix, 2010; Sobek, 2010; Thies, 2010), weak legitimacy (Clapham, 1996; Englebert, 2000), regime characteristics (see Hegre, Ellingsen, Gates, & Gleditsch, 2001) or a breakdown of order (Reno, 2011). Increasingly, research aims to integrate subnational practices, acknowledge the state as a non-neutral actor, and ethno-regional groups as active political agents (Cederman, Min, & Wimmer, 2010). These advancements provide the base for a robust explanation of the changing patterns of violence within African states and the continent as a whole.

Regimes position ethno-regional groups within a power hierarchy and, in doing so, create relative and absolute differences in political authority, capacity, and agency. These 'political inequalities' are formalized through inclusive or exclusive policies. Variable governance relationships produce a spatial landscape of power, and the parameters of each relationship create incentives and disincentives for particular forms of spatially distinct political violence to emerge across that landscape. For example, a discriminated and powerless large group may choose to initiate a civil war in order to usurp power from a regime; however in the same state, agents or allies of the regime may want to limit an opponent by producing targeted and strategic violence through a militia. An underlying assumption is that groups engage in a conflict form that returns the greatest benefits, and maximizes power gains while bounded by their political, economic and social circumstances. This is a robust explanation for multiple, concurrent conflict types compared to subnational 'horizontal' factors including wealth, poverty, ethnic group size, and 'geographic' factors incorporating resource abundance or scarcity.

This idea borrows from the 'topographies' of governance literature (see Boone, 2003; Gupta & Ferguson, 1992) in an effort to address theories of local differences in power, development, autonomy and conflict. It so builds on the emerging, disaggregated 'geography of conflict' literature, which shows clear differences in conflict and non-conflict spaces, and suggests differences in actors, characteristics, and triggers (Cederman & Girardin, 2007; Cederman, Min, & Wimmer, 2010). It seeks to broaden these strands of research by looking at the dominant, common forms of conflict, while incorporating local political dynamics as an explanation for variation. In contributing to these literature, it serves as a basis for an alternative theory of conflict that emphasizes the agency of communities, the importance of scale, the variation in violence, and the consequences of local governance practices.

This article proceeds as follows: the second section reviews theoretical frameworks on subnational governance and conflict across Africa and suggests hypotheses based on positions within ethno-regional hierarchies. The third section is a review of the recently available data on governance and conflict. The fourth section tests the hypotheses and finds support for several; the fifth section discusses and concludes the study.

Subnational governance variation and conflict

Governance is an inherently spatial process, where the impact of inconsistent capacity and control causes subnational variation in government practice, regional integration, development and peace. Subnational theories of conflict argue that violent spaces differ from peaceful areas, and those differences across locations are manifestations of an underlying governance process. Three frameworks explain conflict as a function of governance, yet differ in which mechanisms create subnational variation in authority and the incentives and disincentives for violence. The 'control vs. contest' narrative largely focuses on how poverty and physical geography conspire to limit the effective reach of the state and create opportunities for potential opponents; the 'horizontal inequality' literature considers how political and economic inclusion and exclusion are manifestations of government interest and power across states. Resulting marginalization motivates opponents. The 'hierarchy and landscapes' perspective acknowledges government power and stability as functions of how effectively various politically relevant ethno-regional communities are incorporated, restricted or repressed over space. Each details how a 'topography' of violence results from governance, or lack thereof.

Control vs. contest

The subnational geography literature rests on the notion that the physical and demographic attributes of states create both pools and vacuums of power, where geography creates interplays between strategy, strength and opportunity.

The 'control' narrative argues that states are limited by poverty and the physical challenges of large states, low infrastructure, and difficult terrain. In turn, low state capacity corresponds to more opportunities for groups to rebel and motivation to contest an absent ruling power. Conflicts therefore should occur in rural, peripheral areas or those with rough terrain. Herbst (2000) argues that states use their limited resources to control only areas of demographic and economic significance, Fearon and Laitin (2003) posit that large 'ungoverned', inaccessible areas are more likely to experience civil war, and also serve as a basis for separatist violence. Multiple authors suggest that terrain influences the likelihood of insurgency or that state size creates higher risk: Green (2012) notes that because African states are, on average, 2.4 times larger than European states, their innate risk of conflict is higher (see Alesina & Spolaore, 2003). Alternative manifestations of the 'control' thesis correlate conflict with pre-colonial political organization, colonialist design and post-colonial orders (Crowder, 1968; Herbst, 2000; Young, 1994).

A counter to the control narrative suggests the opposite conditions lead to violence: large states have more violence, but not at a per-capita rate (Raleigh & Hegre, 2009), and areas with difficult terrain are less likely to experience civil war battles compared to accessible areas (Buhaug & R0d, 2005; Raleigh, 2010b; Raleigh & Hegre, 2009). Populated, high value target locations (e.g. resource zones) experience more conflict as maximizing support and resources to allow for continued opposition (Le Billon, 2001). Conflict arises due to contests for power, where opposition forces are sufficiently strong to counter state military power.

'Control' and 'contest' frameworks often disaggregate conflicts to 'separatist' and 'revolutionary' civil wars5 (Buhaug & R0d, 2005), and are largely restricted to determining static contexts in which conflict is more common. If either explanation is uniformly accurate, then conflicts should occur in fixed geographic locations determined by state capacity, physical characteristics, and potential collective action abilities which may benefit nascent groups (Kalyvas & Kocher, 2007). But violence forms and locations vary

widely within and across states, emerging in isolated areas and villages, regional centers, and capital cities.

Absolute and relative inequalities

The 'inequality' framework suggests that states are not neutral agents in their expression of power and privilege, but create access to public and private goods on a 'club' basis. In African states, the included/excluded dynamics are associated with ethno-regional communities, as the political environment is strongly structured along ethnicized patron and client institutions (Chabal & Daloz, 1999; Stewart, 2008). Governments ensure political survival by maintaining the provision of private and public goods to their supportive bases.

'Horizontal inequalities' result from governing policies where 'excluded' groups experience relative deprivation, in contrast to the favored position of 'included' communities (for a critique, see Kasara, 2007). Exclusion along ethnic lines leads to limited representation in public offices (Bangura, 2006), poorer levels of health and education, greater income inequalities (Barron, 2008; Stewart, 2008), and limited public good provision (La Porta, Lopez-de-Silanes, Shleifer, & Vishny, 1999).

Active 'grievances' connect political and economic horizontal inequalities with intra and inter-state variation in subnational conflict. A positive relationship exists between increased conflict and the presence and depth of economic inequality experienced by ethnic or religious identity groups (Buhaug, Gleditsch, Holtemann, 0stby, & Tollefsen, 2011; 0stby, 2008a, 2008b; Stewart, 2008). Further, poverty compounded with social or political inequality leads to higher violence rates (Stewart, Brown, & Langer, 2008). Spatial economic inequalities by group or region are increasingly important in explaining the onset of violence: areas with higher rates of regional poverty are considered more conflict prone, as are spaces of high inequality in heterogeneous societies (Buhaug et al. 2011; Cederman, Wiedmann, & Gleditsch, 2011; Deiwiks, Cederman, & Skrede Gleditsch, 2012; Gubler & Selway, 2012; Murshed & Gates, 2005; 0stby, Nordas, & R0d, 2009).

Yet, economic marginalization may be neither sufficient nor necessary for explaining subnational violence. Increasingly, the differential provision of higher rates of power and public goods access is associated with violence. Stewart (2008) maintains that many horizontal inequalities are tolerated provided political inclusion is widespread; Cederman et al. (2011) find that regardless of relative wealth, discrimination and powerlessness produce higher rates of ethno-nationalist conflict.

Horizontal inequality frameworks propose that the poor and politically marginalized are most likely to revolt, if mobilization conditions are met. Yet this conclusion is difficult to rectify with the known distribution of poverty and conflict risk and the necessary conditions for rebellion. Despite legitimate grievances against a national government, the poorest communities within states do not launch rebel groups, as they often do not have the opportunities (e.g. funds, resources, support) to do so and are unlikely to create the alliances necessary to gather sufficient strength to challenge governments (Collier & Hoeffler, 2002; 0stby & de Soysa, 2008). Instead, the poorest groups are likely to be the most politically disenfranchised (Cramer, 2003) and those outside of the political system are more likely to engage in local contests with other local groups of similar size and stature (Raleigh, 2010a).

Outside of civil wars and related 'ethnonationalist' and secessionist conflicts, how inequality and marginalization create incentives and disincentives for alternative and multiple forms of violence is left unaddressed in the horizontal inequality literature. Indeed, relatively static economic factors do not have sufficient flexibility to explain multiple, different forms of conflict, or propose

conditions under which groups may change the form of violence in which they engage. If all violence did occur in the same relative spaces (e.g. poor areas), the determining factor would be the collective action capabilities of the rebelling group, and would not speak to the comparative advantages different groups have to engage in political violence.

Political hierarchies and landscapes

A third variation on subnational rule argues that governments, vertical power structures and politicized communities can explain both the variation and type of subnational conflict. This literature maintains that regimes rule according to a complex calculus of balancing power interests, and are dependent on a range of subnational alliances, allegiances, and exclusionary policies to bolster stability (Boone, 1998; Chabal & Daloz, 1999); in turn, these create 'governance topographies' (Boone, 2003). Positions within political hierarchies are associated with distinct opportunities and motivations for particular forms of conflict. The spatial clustering of politicized ethno-regional groups creates landscapes of concurrent conflict.

Previous political hierarchies position African ethno-regional groups into static categories of political 'relevance' and 'irrelevance' (Posner, 2004), determined by demography, economic power, national history and positions within past and present governments. Politically relevant (PR) groups are large, live in economically successful areas, and have a substantial role in national politics, while politically irrelevant (PIR) groups include those are that historically marginalized, demographically small, geographically disparate and largely considered the poorest and most vulnerable in society. The political contests of PIR groups are largely local in nature.6

Alternative hierarchies distinguish ethno-regional groups by how they aggregate for political contests. Scarritt and Mozaffar (1999) create national, regional and local 'relevance' categories wherein groups choose alternate political identities depending on the scale of the political issue at hand. Scale determines the identity that allows a group to maximize power relations and minimize the size of issue coalitions.7 This interpretation of relevance suggests that the political aspirations of groups are bounded, but alliances across communities and issues can create strategic opportunities. Yet, the scaled 'relevance' perspective retains a static interpretation of power between group and regime and across groups.

A dynamic, spatial expression of power is necessary to explain variation in positions and political violence. The topography of governance framework builds upon ethno-regional politicization and hierarchical government-group relationships to argue that regimes engage in co-option, alliances, allegiances, negotiations, repressions and non-incorporation across communities to maximize stability and longevity (Boone, 2003; Thies, 2009; Tilly, 1985). This produces a pyramid of power relations between the government and ethno-regional communities (Boone, 2003; Gupta & Ferguson, 1992), where a group's position reflects how central it is to the composition and legitimacy of the ruling regime. These positions are also relative to other ethno-political groups who are contending for access to executive power.

The logic behind this interpretation of governance is that leaders of heterogeneous states have multiple potential rivals to placate and incorporate, rather than simply repress. A hierarchical framework allows access to power for the select groups deemed to be most critical in maximizing regime scope and longevity, while excluding groups who may threaten it. Across a state at any time, those groups in power or allied to those in power are juxtaposed with excluded groups who occupy positions of powerlessness, discrimination, and autonomy. Hence, positions within a political

hierarchy are determined by whether groups are integrated into national politics by constituting part of the regime, or remain external through repression, discrimination, or irrelevance. Both are situated above those with little or no political strength or relevance (i.e. politically irrelevant) (Cederman et al. 2010). Fig. 1 is an illustration of the political hierarchy.

The contours of subnational governance relationships change as regimes consolidate power while renegotiating and restructuring political allegiances with subnational elites to secure stability (Boone, 1998). As regimes change, these hierarchies adjust to reflect the supportive communities and alliances of new dominant ethno-regional leaders or regimes. Ongoing and multiple political changes produce dynamic conflict landscapes across states, in which groups and governments face a shifting rate of opposition, capabilities, motivations and types of conflict.

The spatial implications are similar to alternative subnational frames, wherein political weight creates spatial variations of power and influence. Power pools in 'included' areas, and is largely absent from others, depending on the position of local ethno-regional groups within political hierarchies. The evident variation therein creates a landscape of political power across that state as ethnic communities across Africa are spatially clustered and can be identified within a specific homeland (Scarritt & Mcmillian, 1995).8

Explaining violence occurrence and types

Governance practices that produce power hierarchies are based on local negotiation, incorporation and repression, and the variable political relationships produce a landscape of power. This explanation allows for different comparative advantages for groups to orchestrate conflict across a state, but also engenders specific motivations for violence as groups seek to retain, change or usurp power, while bounded by their abilities. If the locations in which distinct conflicts occur differ according to heterogeneous political status, this is evidence that conflict forms are endogenous to group relationships with governments, and that the process of governance leads to different and multiple forms of conflict, each according to groups' means and goals. To relate this theory of landscape and political hierarchies to violence types, each conflict form is associated with the likely underlying relationship.

Rebel-based civil wars

A civil war's dominant cleavage involves a rebel group battling against government forces, with the objective of replacing the national regime or establishing a separate state (Gleditsch et al. 2002; Sambanis, 2004). Rebel groups are political organizations designed to counter an established governing regime through violence; they have a stated political agenda and the group is acknowledged outside of immediate members. Civil wars are more likely in countries where groups are marginalized and excluded from power by a repressive government (Cederman et al. 2011).

Groups most likely to engage in civil wars are politically relevant communities who are not presently benefitting from privileged

Dominant/Monopoly |

Partners (Senior & Junior) |

Powerless |

Discriminated |

Politically Irrelevant

Most Included in Regime

National Power Scale

Least Included in Regime Fig. 1. Political hierarchy.

positions in a power hierarchy. Political relevance allows for collective action and the assembly of a significant threat, and marginalization and/or discrimination relative to other communities serve as motivations. Therefore, civil wars are a form on interelite violence: as rebelling group(s) need to garner significant support to challenge an existing political network, a contest of power between similarly strong and relevant groups is most likely.9 The 'contest' dynamic between large groups is evident in several recent examples, including the struggle over the Congolese Kivus (Lemarchand, 2009), Ivory Coast (Boone, 2007), Congo-Brazzaville (Bazenguissa-Ganga, 1999; Lemarchand, 2009), Chad and the Central African Republic (Giroux, Lanz, & Sguaitamatti, 2009).

Hence, civil war events are most likely to be pursued by both 'powerless' and 'discriminated' politically relevant groups. Relative poverty levels will not affect the likelihood of who rebels, as those are endogenous to positions within the political hierarchy. This explanation is largely in line with assumptions of contemporary literature on civil war that gives particular significance to ethnic exclusion and the resulting "grievances" as major sources of civil war (Cederman et al., 2010, 2011; Gurr, 1993, 2000; 0stby et al., 2009; Stewart, 2008).

Hypothesis 1. Discriminated and powerless groups are more likely to engage in rebel-based civil wars over other forms of conflict.

Political militia violence

Political militia violence is orientated towards shaping the political power structure at the local, regional and national level. This violence has increased as a result of political fragmentation and competition (Branch and Cheeseman, 2009; Bratton & Chang, 2006; Chabal & Daloz, 1999), and in response to the rising level of checks and balances on the executive power (Choi & Raleigh, 2014). Power distribution across parliaments, judiciaries and the military, decentralization, and election monitoring increase the number and power of non-regime elites shaping government policy. In turn, this leads to high levels of elite competition and fragmentation over access to state resources and power across African states in particular (Brancati, 2007; Lijphart, 1977). Within systems where political office has redistributive implications, incumbents and opponents have incentives to design forms of violence to assure access to power (Arriola & Johnson, 2012; Gandhi & Lust-Okar, 2009; Schedler, 2006; Treisman, 2007). Hence, using violence to secure a position in government is an effective tool of competing elites (including members of government) where the stakes are exceedingly high. The manifestation of such competition is in the use of militias as personal armies for politicians (e.g. Somali regional militias), to challenge voters or oppose candidates during election periods (e.g. Mungiki in Kenya), and 'unaffiliated' pro-government supplementary squads (e.g. Janjaweed and PDF in Sudan).

Fig. 2. Political relationships and conflict spaces.

Militias are the violent group choice for elites who aim to capture or retain power through the strategic and orchestrated use of violence by a private army10. In choosing this form of conflict, those in power seek to redress perceived power imbalances through force: challengers seek to shift positions within a governance hierarchy, not to orchestrate a national overthrow or redress local political disputes. Militia activity is a form of intra-elite violence, used by those within political hierarchies to threaten, force or impede political changes (e.g. party militias, election gangs).

Therefore, it follows that militia activity is by, and for, 'included' elites, who aim to capture or retain power through the strategic and orchestrated use of violence.

Hypothesis 2. Those included within regime hierarchies, such as partners, dominant and monopoly groups, are more likely to use political militia tactics over other forms of violence.

Communal conflict

Communal conflict largely occurs in geographically 'peripheral' spaces, and is strongly associated with specific, local 'security' providers who are members of ethnically distinct long term policing units, such as those common among Somali clans, or protector/predator groups engaged in raiding or traditional retribution violence (e.g. Kenyan Pokot Militia). Communal episodes are typically brief, often deadly, traditionally based contests for local level resources and micro political dominance. Participation is strongly linked to local identities, and groups often engage in violence that takes on the attributes of 'livelihood defense' as the catalysts tend to be land, water, grazing rights, cattle etc. Informally organized religious militias (e.g. Christian and Muslim attacks) also fall under this category.

Very few episodes of communal violence involve police or military forces; most activity is directed towards other communal groups and or civilians from opposing communities. These power struggles are locally bound and mainly present in areas deemed politically irrelevant, as they are populated by small groups which do not aggregate into communities of regional and national political relevance. As research suggests that these groups are often peripheral to national governments as they have little 'political weight' (Raleigh, 2010a) and are effectively excluded from the political hierarchy, local political conflict is most likely.

Hypothesis 3. Politically irrelevant communities resort to communal violence at higher levels over other forms of political contest.

To summarize: the political hierarchies and landscape framework proposes that groups are constrained by their relative positions within national power hierarchies. Positions determine conflict goals, opponents, and threat potential to regimes. Divergence in conflict type is due to government inclusion processes and group abilities. Distinct violent spaces should therefore consistently differ by governing relationship: civil wars should predominately occur in spaces where the government practices discriminatory and exclusionary practices; militia actions will mainly take place in areas that are included and integrated into the government structure; and communal violence will largely occur in zones that are 'ungoverned' and unincorporated into the national regime. In turn, this influences the risk across groups: powerless and discriminated groups, relative to other groups in a political hierarchy, choose a more fatal and long term form of conflict; powerful groups engage with other elites through creating short periods of high violence and targeted fatalities; and small, local ethnic groups engage in high levels of contained violence. Ignoring the complex, scaled and interactive domestic politics de-politicizes conflict and mis-

identifies cause and triggers.11 See Fig. 2for a representation of multiple forms of violence within a hypothetical state: instead of a uniform and generic form of homogenous political instability in country A, multiple forms can emerge due to local relationships.


The motivations and goals of challengers are altered by changes in relative power, such that advancements or regressions on a political hierarchy may heighten grievances of civilians, elites or repressed politically relevant communities. Therefore, as a group changes position on the political hierarchy, its dominant form of conflict should also adjust. A more inclusive position on the power hierarchy is the main mechanism for the transition between civil war and militia activity as the dominant form of political violence; a less inclusive position on the same hierarchy explains the movement from militia to civil war violence.

Hypothesis 4a. : A change to a more inclusive position on a political hierarchy will correspond to a change from civil war to political militia activity across localities.

4b. : A change to a more exclusive position on a political hierarchy will correspond to a change from political militia to civil war activity across localities.

Finally, national institutional systems may be a critical facet of the governance equation as they shape and contain the abilities and motives of central regimes, and possibly reflect how certain types of conflict become more likely: autocratic and transition regimes remain open to civil wars due to their limits to inclusion, and narrow political hierarchy (Hegre et al., 2001). In contrast, democratic institutions force governments to engage with more 'partners', as regimes can no longer build narrow and exclusive coalitions to pursue their policies and protect themselves. This leads to fewer motives for civil war as the paths to political change are more open and permeable, yet democracy produces both competition and fragmentation amongst political elites on multiple scales (Branch & Cheesman, 2009; Berman, 1998; Mozaffer, Scarritt, & Galaich, 2003); political militias are more active in democratizing states and particularly in periods of competition, such as elections. Further, as states are forced to democratize, there are changes in how different groups are integrated (e.g. a higher rate of inclusive coalitions is necessary in all states excepting those with a winning ethnic majority); hence more or less inclusion and exclusion is related to regime type. Yet, there is little in the institutional literature that associates variation in regimes with subnational incentives for the rise of multiple and alternative conflicts. Previous work into how regimes and conflict types change in tandem

0.015 0.01 0.005 0

Irrelevant Excluded Included

Fig. 3. Civil war risk box plot.

suggested that a dominant form arises as political participation and government incentives for social freedoms assume a more central position in a polity (Tilly, 1985). Hence, regime perspectives cannot answer subnational queries on concurrence and the lack of spatial overlap.

Research design

This study uses grid-years as the unit of analysis. The complete dataset covers Sub-Saharan African states from 1997 to 2011 with populations over 100,000, wherein each state is divided into approximately 10 km x 10 km units, repeated by year.12 3,800,940 grid-year units are uniquely identified, and further clustered into administrative zones (1st, 2nd, and 3rd where available), ethnic communities, and countries. Conflict point data is grafted onto this grid structure. The sub-national level of analysis is particularly appropriate in contexts where a particular form or agent of violence is geographically concentrated. Grid units have the additional advantage of being disassociated from formal political boundaries or units of administration, and allow for a robust test of local characteristics that make locations more or less conflict prone.

Conflict data are from ACLED's 60,000 disaggregated African event points from 1997 to 2011 (Raleigh et al., 2010). ACLED has an 'atomic' event format, where each event is coded as a daily, georeferenced, actor-specific occurrence, allowing for the greatest possibilities for aggregation and comparison (e.g. all 'rebel' actions or singular group actions; all events by month, etc.), as all event units are the same across time periods and countries. Further, ACLED codes violence systematically across several state and nonstate actor types without a fatality-based criterion or pre-defined conflict distinctions.13 Georeferenced conflict points are first categorized into distinct 'types' (rebel, militia, communal, other), based on actor goals and self-proclaimed designations; each point is then associated with the specific grid-year. All conflict units are associated with a buffer that controls for a slightly larger area and incorporates events that occur in grid border regions; the point is assigned to the grid in which the majority of the buffered point lies.

There are three different forms of the dependent variable: 1) an event count of each type; 2) a dominant conflict type; and 3) a change in dominant conflict. A 'dominant' count variable records the most active type of conflict in each grid-year, and the number of events therein. This count controls for singular events by other types of actors and produces an accurate interpretation of where distinct types of conflicts occur. Three dichotomous variables of conflict type are based on dominant counts: each grid-year with a dominant rebel count is coded '1 '; this is repeated for dominant militia and communal event counts. Conflict in the dataset is distributed as follows: civil war (1.53%), political militia activity (1.65%), communal conflict (.62%), and no conflict (96.1%).

The dichotomous 'change' variables are based on whether the dominant form of conflict changes during the previous year: militia change is coded '1 ' if a country experiences an 'inclusive' transition from civil war to political militia; rebel change is coded '1 ' if 'exclusive' when moving from political militia to civil war.

Governance relationships are derived from the Ethnic Relations Dataset (EPR) (Cederman et al., 2010; Wimmer, Cederman, & Min, 2009) which provides time varying information on the political status of ethno-regional groups. Associating each grid square with a local ethnic community involved a three stage process: first, a complete georeferenced list of groups is compiled and mapped; second, each group is designated as politically relevant or irrelevant; and third, the changing governance relationship is ascribed to each group by year. Creating a complete list of spatially referenced groups involved working across three ethnic data sources for Africa. The names and locations of groups in the 'Ethnic Power Relations'

Table 2

Summary variables.

Variable name and time coverage



Conflict 1997-2011

Infrastructure Population

Urban centers

Higher than mean wealth

Ethnic power and relevance



Atomic events by agent, date and location, aggregated

into grid squares. A lag variable for previous conflict

and spatial summary variable records the number

of conflict events in surrounding locations (logged

to control for high counts).

Civil war: .069 (mean) 1.32 (SD) 0-355 (Range)

Civil war dummy: 1.4% Positive 98.6% Negative

Political militia: .053 (mean) 1.25(SD) 0-460 (range)

Political militia dummy: 3.4% Positive 98.6% Negative

Communal violence: .014 (mean) .328(SD) 0-61 (range)

Communal violence dummy: .46% positive 99.54% negative

Length of roads per grid square

26.21 (mean) 45.97 (SD) 0-1171 (range)

2.5 min grid cell population density data


Mean: 3215

SD: 11,588


Capital city, administrative capitals, cities (over 50,000)

per grid square. Quartile measures summarized

The mean annual rate of income for the state, a dummy

variable indicating whether a grid square was above the mean

Mean dummy: 26.67% positive 73.33% negative

Wealthy grid dummy: 16.21% Positive 83.79 negative

Each ethnic group is designated annually as dominant,

senior partners, junior partners, powerless, discriminated or irrelevant.

Politically irrelevant: 27.3% positive 72.7% negative

Discriminated: 3.13% positive 96.87% negative

Powerless: 21.93% positive 78.07% negative

Partners: 29.71% positive 70.29% negative

Dominant/Monopoly: 17.81% positive 82.19% negative

POLITY dataset

-66, -77, and -88 converted to '0'.

All instances over +5 are coded as consolidated democracies

Democracy: 33.7% positive 66.3% negative

Dummy variables constructed based on the occurrence of an

election (presidential or parliamentary) in a given year

Election years: 22.4% of sample

Gross domestic product, in 2000 terms


23.06 (mean) 1.42 (SD) 19.04-25.95 (range)

Local georeferenced to village and counted by daily occurrence

Local (10.8 km grid)

Local (in decimal degrees)

250 km Grids 1995, 2000, 2005.

Ethnic Group-years




FAO: Roads of the World

ESRI: African

UNEP/GRID African population distribution

Ciesin Gecon data

(a) Ethnologue

(b) Scarritt & Mozaffar, 1999; Joshua Project

(c) Ethnic Power Relations Dataset


African elections database

World Bank

(and associated GEO-epr) dataset differ from the consistently updated Ethnologue ethnicity and language map of Africa, which provides a spatially referenced map of group homelands. Further, information on names and relationships of scaled, politically relevant ethno-regional groups by Scarritt and Mozaffar (1999) supplement the apolitical Ethnologue dataset and more general GEO-epr set. GEO-epr provides information for just under half of the continent, and is composed of aggregate group names and regions (223 groups are identified overall). The supplementary disaggregated sources are more similar to each other, report on local and subnational identities, and contain information for over five times as many discrete groups as GEO-epr. The final list and map both classify and distinguish sub-groups into identifiable ethno-regional clusters.

After associating each grid square with an ethno-regional group, an annual political relationship status is assigned. To begin, each group is broadly categorized as politically relevant or irrelevant from Scarrit and Mozaffer's (1999) list. Groups that are 'politically irrelevant' do not appear as named or associated kin groups in both GEO-epr and Scarritt & Mozaffar lists. Political irrelevant groups do not change in their status through the observed time period, as politically relevant groups do. GEO-epr designates groups as 'Included' if from 'Dominant/Monopoly' (indicating dominance of an ethnic community in government) and 'Partners' (both senior

and junior associated groups) categories; 'Excluded' if from 'Discriminated' (ethno-regional communities subject to repression), 'Powerlessness' (ethno-political groups subject to exclusion), or 'Autonomy' categories. Groups who are kin associates of identified 'powerless' groups are also categorized as such, as are those that appear in the Scarritt and Mozaffar politically relevant list, but not in EPR. The dataset has the following distribution: dominant/ monopoly (3.5%), partners (37%), discriminated/powerless (31.5%), and politically irrelevant (28%)14.

Additional characteristics are grafted onto each grid, including regional and local information on infrastructure, demographics, and wealth. Local population and infrastructure are found to be powerful indicators of civil war clustering (Raleigh & Hegre, 2009); further regional and group poverty rates is a key explanation in civil war likelihood (Cederman et al. 2011). A petroleum resource indictor is included as resource wealth is found to positively affect opportunity abilities of groups for conflict (Basedau & Pierskalla, 2014). A standard suite of national control variables in all models includes the natural log of a country's GDP, and the year of Presidential Election and whether the state is a qualified democracy (Polity scores of or greater than +5). A presidential election dummy variable controls for the effects of electoral competition, and democracy for institutional governing systems. Each included variable has been affirmed as significant controls in econometric analysis on

Table 3

Logit analysis of dominant conflict form.

Model 1* Model 2* Model 3*

Civil war dominance Political militia dominance Communal violence dominance

Political power hierarchy

Powerless/Discriminated .367(.185)** .139(.172) —.434(.203)**

Partners (Sn & Jr) .327(.258) .797(.337)** .249(.469)

Dominance .914(.696) .956(.540)* —.931(.760)

Over mean wealth .336(.118)** .405(.114)** .092(.221)

Presidential election —.376(.267) .328(.238) —.134(.278)

Local log population .183(.097)** .405(.052)*** .340(.070)***

Distance to cities —.234(.106)** —.418(.175)** —.124(.151)

Admin capital .273(.112)** .192(.189) .285(.256)

Infrastructure .004(.000)*** .004(.000)*** .005(.000)***

Log (GDP) —.203(.136) .034(.144) .150(.239)

Democracy —.879(.333)** —.510(.333)* —.315(.439)

Petro area 1.32(.486)** 1.48(.254)*** 1.64(.349)

Lagged conflict typet_1 1.57(.176)*** 1.99(.121)*** 1.74(.123)***

Additional civil war lagt_2 .38(.136)**

Surrounding violence (Admin 2) .822(.053)*** .627(.071)*** .710(.079)***

Constant — 1.99(3.40) —8.63(3.19)** — 11.75(5.24)

Log-likelihood — 190,706 —219,526 — 112,341

Rho .26 .224 .16

Observations 3,011,910 3,011,910 3,011,910

Countries 42 42 42

(Notes. Standard errors in parenthesis.*p < .1; **p < .05; ***p < .01. Two-tailed tests).

conflict (Sambanis, 2004). Table 2 provides summary statistics for all independent variables.

A lagged dependent variable adjusts for autocorrelation within clusters in the time-varying dataset (Wilson & Butler, 2007); in the case of civil war, a two-year lag is included as the rate of correlation rises until two years have passed and decreases substantially thereafter; a yearly lag variable of each discrete type of conflict, as well as a summary count of surrounding violence in 50 km and 100 km zones. The 100 km district space is chosen as the conflict spatial dependence control.

Two statistical models test the hypotheses: a negative binomial model for event count dependent variables and a standard logit model for binary dependent variables of dominant conflict form and change within grid-years. Robust standard errors are clustered by country.


The results for both logit and count models confirm that there are clear differences in the spaces of rebel-based civil wars, political militia activity, and communal violence. These types of violence are spatially distinct and concurrent. Table 3 presents logit results for

the three types of conflict: Model 1 confirms that the relative position in a political hierarchy is a strong explanation for conflict type: areas characterized by political powerlessness and discrimination experience more rebel events than those in partnership with regimes or with a monopoly on power. The immediate past and present conflict environment supplements risks presented by local physical and demographic geography: populated, accessible zones that are conflict active in the present and near past experience an additional risk. Relative wealth exerts a limited positive influence on the occurrence of civil war events. As predicted by the theory and hypotheses 1, rebellion is more likely to occur in areas excluded from the regime where competition between regime and opposition elites is highest. Fig. 3 displays a risk box plot for civil war rebel violence where the mean rate and range of violence in 'excluded' areas is up to twice that of 'included' or 'irrelevant' areas.

In model 2, political militia violence is most common in areas characterized by regime inclusion, whether in the form of junior and senior regime partnerships, or regime dominance. Both 'partnered' and 'dominant' regions are twice as likely to experience political militia violence over 'excluded' positions on the political hierarchy. This supports hypothesis 2, and is robust when controlling for the presence of a civil war within a state (to reduce the

0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0




Fig. 4. Political militia risk box plot.

Irrelevant Excluded Included

Fig. 5. Communal violence risk box plot.

effect of government or rebel groups using militias during conflict) and without such controls. These results are consistent with the theoretical expectation that competition and fragmentation amongst included elites is a primary determinant of militia violence. Overall, the differences in where militia violence and civil war events occur largely differ on how governments engage with the populations, yet the economic and demographic characteristics of environments experiencing both are similarly wealthier, resource abundant, and accessible parts of states. Fig. 4 displays a risk box plot for political militia violence where the mean rate and range of additional violence in 'included' areas are more than three times the rate across other political categorizations.

In model 3, communal violence is less likely to occur in those areas characterized by discriminated, powerless, and dominant political positions. Powerless areas return a significant, negative sign compared to politically irrelevant places (the reference group); partnered and dominant areas are insignificant. Unlike the other forms of armed violence, relative wealth is not a factor in areas with higher rates of communal violence. Previously violent, resource wealthy, accessible, rural areas are more likely to have a communal event. National characteristics are not strong predictors of communal violence, although fewer civil wars are found in democratic states, and militia violence weakly corresponds. Fig. 5 displays a risk box plot for communal violence where the mean rate and range is up to four times the violence risk in other political areas.

Table 4 presents the negative binomial model results which largely confirm the conclusions of the logit models, but with the following differences: in the count models for civil wars, higher event counts also occur in dominant areas. This is largely in line with expectations that civil wars eventually descend upon capital cities, and into the territory of the powerful (Raleigh & Hegre, 2009). Militia event frequency increases during election periods, and in areas of higher than mean wealth. All other key variables results relating to ethno-political hierarchies and relative income are stable and confirmed.

Table 5 presents logit results for change. For comparison and robustness checks, models 7 and 8 show results with the same set of independent variables used in models 1 -6, but the dependent

Table 4

Negative binominal analysis of event counts.

Model 4 Model 5 Model 6

Civil war Political militia Communal violence

Political power hierarchy

Powerless & Discriminated .527(.208)** .104(.149) —.595(.190)**

Partners (Sn & Jr) .348(.272) .900(.316)** .129(.389)

Dominance & Monopoly 1.41(.512)** 1.11(.533)** —1.13(.788)

Over mean wealth .166(.153) .283(.102)** -.092(.162)

Presidential election -.526(.223f* .353(.253) -.331(.222)

Local log population .217(.106)** .427(.053)*** .360(.057)***

Distance to cities -.352(.128)** -.448(.148)** -.195(.127)

Admin capital .409(.128)** .311(.188)* .428(.238)*

Infrastructure .004(.000)*** .005(.000)*** .006(.001)***

Log (GDP) -.419(.134)** -.056(.150) .155(.203)

Democracy -1.11(.299)*** -.497(.299)* -.503(.422)

Petro area 1.79(.570)** 1.73(.267)*** 1.94(.378)***

Lagged conflict typet-1 1.21(.102)*** 2.48(.131)*** 2.01(.159)***

Additional civil war lagt-2 1.77(.130)***

Surrounding .705(.124)*** .882(.107)*** 1.04(.156)***

violence (Admin 2)

Constant -4.59(3.19) -14.16(3.26)*** -19.12(4.47)**

Ln (year) 1 1 1

Log n alpha 3.14 2.89 3.99

Observations 3,011,910 3,011,909 3,011,910

Countries 42 42 42

(Notes. Standard errors in parenthesis.*p < .1; **p < .05; ***p < .01. Two-tailed tests).

Table 5

Change logit analysis of dominant conflict form.

Change from militia to civil war dominance Change from civil war to militia dominance

Change to included Status .538(.309)*

Change to excluded status .599(.335)*

Local log population .370(.053)*** .416(.051)***

Distance to cities -.489(.143)** -.456(.146)**

Admin capital .187(.157) .244(.161)

Infrastructure .004(.000)** .004(.001)***

Log (GDP) .013(.124) .053(.140)

Elections -.329(.254) .433(.215)**

Democracy -.346(.253) -.346(.256)

Over mean wealth .114(.142) .184(.132)

Petro area 1.09(.224)*** .105(.212)**

Surrounding violence .275(.048)*** .549(.051)***

Constant -6.92(2.89)** -8.72(3.22)**

Log-likelihood -169,647 -164,732

R2 7% 12%

Observations 3,011,910 164,732

Countries 42 42

(Notes. Standard errors in parenthesis.*p < .1; **p < .05; ***p < .01. Two-tailed tests) Lagged violence.

variables are whether the dominant conflict in a grid-year differed from that in previous years (lags of 2-3 years). The shifts are either from (1) dominant militia events to civil war or (2) dominant civil war events to militia. The table shows that a higher rate of inclusion within political hierarchies is a conclusive explanation for shifts between civil war and militia events, and that a transition from an included to an excluded status on a political hierarchy is associated with a shift in violence from militia to civil war events. Hypothesis 4 is therefore accepted, as these results indicate that groups have agency to change the form of political violence they engage in when the political parameters shift and new motivations, opportunities and goals emerge. However, it is not an immediate change, as 2-3 years are necessary for groups to organize and motivate followers for alternative forms of conflict goals.

The results for change between militia and civil war show minor differences in the effect of key independent variables: both suggest that positions on political hierarchies, concurrent violence and area characteristics are significant indicators. By and large, concurrent temporal shifts are less relevant, barring the significance of elections in motivating increased militia violence.

In summary, hypotheses 1 -4 are fully or partially confirmed; in count models, civil war events occur in large number in dominant areas, although there is less risk of civil war events in 'included' positions of the political hierarchy; rebel events are a form of interelite competition between included and excluded politically relevant ethno-regional groups. Political militia events largely occur in areas characterized by regime inclusion; hence this form of political violence is an intra-elite form of competition. Communal violence is most common in areas where politically irrelevant groups reside, and less likely to occur in areas characterized by regime inclusion or exclusion. Finally, as political relationships change, so too do conflict forms; this is evident in the shifts in militia to civil war events across space. Each analysis is robust to several alternations to the dependent variable, the main independent variables, stepwise deletions and bootstrapping.

Discussion and conclusion

The topography of political power across contemporary African states creates variable forms and spaces of conflict. This analysis confirms that political hierarchies and power landscapes are strong explanations for spatially distinct, concurrent and changing types

of political violence. The conclusions are in line with recent studies indicating that ethnic exclusion and the resulting grievances are major sources of civil war (Cederman et al. 2011; 0stby et al., 2009). Yet, civil wars are only one form of violence within the larger conflict landscape that emerges across Africa as a response to governance: while discriminated and powerless groups choose civil war, other groups within the same state oppose regimes in alternate ways that maximize their abilities. These groups consider the incentives and opportunities provided by their position to shape their struggle within the political environment. Concurrent forms of internal political violence differ in practice, scale, goals, underlying political context and consequences.

Conflict literature on African states generally presents a poor conceptualization of power, and a myopic understanding of the production and manifestation of all political violence. This study asserts that political violence within African states is a response to, and shaped by, the type of governance practiced by political elites; the evident uneven development and political inclusion within and across African states creates a landscape of variable conflict types and risks.

From this analysis, inclusion and marginalization cannot fully account for the range and occurrence of violence. Regimes create political hierarchies to organize how they will be represented across territory: they incorporate allies, integrate potential regional rivals, placate other powerful elites and repress threatening groups. Positions within that hierarchy and landscape shape how groups challenge sitting governments, the form that opposition takes, and the conditions under which it will be successful. This hierarchy can be understood spatially as a political landscape, where group homelands are associated with different levels of political power, dominant conflict types and potential. Other frameworks designed to explain disaggregated conflict suffer from an almost exclusive focus on civil war, a lack of information on how regimes create incentives and disincentives for political violence, and a dismissal of group agency and reactions to changes in political environments. The political hierarchy framework integrates explanations of horizontal and vertical inequalities to produce dynamic and variable explanations. It is a direct counter to reactive and static interpretations of conflict geography, risks and agents.

The implications for violence across Sub-Saharan Africa are that states can have complex conflict geographies, involving violence with variable levels of lethality, national security significance and duration. The key factor to consider is at which scale the various contests are being fought: underlying the subnational contexts and the presence of different manifestations of opposition/violence is the political logic and process of governance across African states. The structure and consequences of African governance are critical to conflict researchers seeking to better understand how geopolitical fault lines emerge within a state: local politicians are not simply pawns of national elites, but organize and mobilize to receive public goods to compensate for their support. A breakdown in these relationships can cause national instability. As regimes change, corresponding allies, policies and enemies shift, as does the political geography of power within a state. For the purposes of understanding the creation of violent opposition, the most egregious consequences of ethno-political hierarchies are that they create a shrinking political arena (Kasfir, 1976) and encourage strategic uses of violence to attain power. In practice, regimes create a 'zero-sum' political atmosphere where those outside of favored ethnic status are not afforded representation based on public support, demography or ideology. Those who are excluded can form an opposition for which violence may become a path to power. Further, these relationships are dynamic, conflict types and rates are territorially bounded and not closely associated to total poverty or total marginalization, but dynamic relative power

positions and status. African conflict therefore requires simultaneously temporal, spatial and disaggregated explanations.

But is Africa unique? African states are substantially more underdeveloped and have a higher risk of state failure than other post-colonial countries. This is largely due to the extent and depth of client politics, wherein the regime leaders use public funds for personal discretion and the power structure in the state is largely confined to ethno-regional hierarchies and associations of the power elites. The effects of widespread and frequent conflict on African democratic stability and economic growth are major concerns for the global community: internal political instability reverses development progress (Collier et al., 2003) through weakening governing institutions and creating conflict traps. Such violence has long term and detrimental effects on the welfare and health of citizens through reduced life expectancy (Ghobarah, Huth, & Russett, 2004), continued underdevelopment, unstable democratic transitions (Hegre et al., 2001), weakened livelihoods and state failure (Clapham, 1998). Yet neither democracy nor development are complete solutions for African violence: civil war may be decreasing, but alternate forms of political violence are growing forms of opposition. Without more concerted efforts to understand how governance creates - rather than curtails - conflict, such instability is likely to continue.


The work was supported by the European Research Council Young Investigator Grant: Geographies of Political Violence.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://


1 An exception to this is Le Billon (2001) who captures how resource geographies structure the actions and structure of groups.

2 A politically violent event occurs when one or more groups use force on a specific date and in a location. Particular aggregations of atomic events into other categories (e.g. urban unrest, coups, rebellions etc.) are not determined a-priori within ACLED.

3 These common forms are not exhaustive: political violence includes rioting/protesting, mutinies etc.

4 In Table 1, an 'active' count is created for each form of conflict by calculating the annual country mean number of grid events, then classifying those which equal or surpass that total during the specific year. This reduces the effect of low/singular events.

5 Recent work on climate change-conflict uses the capacity framework for communal and social conflict.

6 Pastoralist groups, clustered into distinct and often-small ethnic communities, are an apt example of the politically irrelevant .

7 For example, a member of the politically relevant 'Anambra' community in Nigeria uses that identity for local political contexts, aggregates with other communities into the 'Igbo' identity in regional contests and is subsumed under the nationally-relevant 'Southern' community in national contests.

8 The assumption underlying each ethnic dataset, and this analysis, is that African ethno-regional groups cluster in space.

9 The 'strength' distinction is relative: a rebelling group needs to be able to sufficiently challenge a seated government, and the strength and capacities of governments vary significantly.

10 Militias are used by multiple types of elites, including governments (pro-government party gangs like Zimbabwe's ZANU-PF) and opposition leaders (pro-opposition party gangs); on occasion, rebel leaders engage militias.

11 Groups with mobilization potential (either by size, support, or history) populate all categories except 'political irrelevance'. Political irrelevance remains a category as small ethno-regional communities are not engaged in national politics, but self-identify as a group and are powerholders on the local level.

12 Both Rwanda and Burundi are excluded from the models, as the ethno-regional geography between main groups is indistinguishable. Additionally, states with populations under 100,000 are excluded because governance, demographic and violence patterns are unique and distinct from larger states.

13 ACLED defines political violence as 'the use of force by a group with a political purpose or motivation'.

14 Sub-Saharan Africa only. Comparisons across ethnic sets are available in the Appendix.

15 Additional results using different ethnic distinctions, variations on the hierarchy variable and samples are available in the Appendix.


Alesina, A., & Spolaore, E. (2003). The size of nations. Cambridge, MA: MIT Press.

Arriola, L., & Johnson, E. (2012). Election violence in democratizing states. Under Review: Berkeley, California: University of California: Berkeley.

Bangura, Y. (2006). Ethnic inequalities and public sector governance. Basingstoke: Palgrave.

Barron, M. (2008). Exclusion and discrimination as sources of inter-ethnic inequality in Peru. IFPRI: Washington, DC.

Basedau, M., & Pierskalla, J. (2014). How ethnicity conditions the effect of oil and gas on civil conflict: a spatial analysis of Africa from 1990 to 2010. Political Geography, 38(1), 1-11.

Bates, R. (2008). When things fell apart: State failure in late-century Africa. Cambridge: Cambridge University Press.

Bazenguissa-Ganga, R. (1999). The spread of political violence in Congo-Brazzaville. African Affairs, 98(390), 37-54.

Berman, B. (1998). Ethnicity, patronage and the African state: the politics of uncivil nationalism. African Affairs, 97(2), 305-341.

Boone, C. (1998). State building in the African countryside: structure and process at the grassroots. Journal of Development Studies, 34,1-31.

Boone, C. (2003). Political topographies of the African state: Territorial authority and institutional choice. Cambridge: Cambridge University Press.

Boone, C. (2007). Africa's new territorial politics: regionalism and the open economy in Cote d'Ivoire. African Studies Review, 50(1), 59-81.

Brancati, D. (2007). Decentralization: fueling the fire or dampening the flames of ethnic conflict and secessionism? International Organization, 60(3), 651-685.

Bratton, M., & Chang, E. (2006). State building and democratization in Africa: forwards, backwards or together? Comparative Political Studies, 39(9), 1059-1083.

Buhaug, H. (2006). Relative capability and rebel objective in civil war. Journal of Peace Research, 43, 691-708.

Buhaug, H., Gleditsch, K. S., Holtemann, H., 0stby, G., & Tollefsen, A. F. (2011). It's the local economy, stupid! Geographic wealth dispersion and conflict outbreak location. Journal of Conflict Resolution, 5(5), 814-840.

Buhaug, H., & Rod, J. K. (2005). Local determinants of African civil wars, 1970-2001. Political Geography, 25(2), 315-335.

Cederman, L. E., & Girardin, L. (2007). Beyond fractionalization: mapping ethnicity onto nationalist insurgencies. American Political Science Review, 101(1), 173-185.

Cederman, L. E., Min, B., & Wimmer, A. (2010). Why do ethnic groups rebel? New data and analysis. World Politics, 62(1), 87-119.

Cederman, L. E., Wiedmann, N., & Gleditsch, K. S. (2011). Horizontal inequalities and ethno-nationalist civil war: a global comparison. American Political Science Review, 105, 478-495.

Chabal, P., & Daloz, J. P. (1999). Africa works: Disorder as political instrument. Oxford: James Currey.

Cheesman, N., & Branch, D. (2009). Democratization, sequencing, and state failure in Africa: lessons from Kenya. African Affairs, 108(430), 1-26.

Choi, H. J., & Raleigh, C. (2014). Dominant forms of conflict across regimes. International Studies Quarterly (forthcoming).

Clapham, C. (1996). Africa and the international system: The politics of state survival. Cambridge: Cambridge University Press.

Collier, P., Elliott, V., Hegre, H., Hoeffler, A., Reynal-Querol, M., & Sambanis, N. (2003). Breaking the conflict trap : Civil war and development policy (Vol. 1). Washington, DC: World Bank Press.

Collier, P., & Hoeffler, A. (2002). On the incidence of civil war in Africa. Journal of Conflict Resolution, 46(1), 13-28.

Collier, P., & Hoeffler, A. (2004). Greed and grievance in civil war. Oxford Economic Papers, 56(2004), 563-595.

Cramer, C. (2003). Does inequality cause conflict? Journal of International Development, 15, 397-412.

Crowder, M. (1968). West Africa under colonial rule. London: Hutchinson.

Deiwiks, C., Cederman, L. E., & Skrede Gleditsch, K. (2012). Inequality and conflict in federations. Journal of Peace Research, 49(2), 289-304.

Dixon, J. (2009). What causes civil wars? Integrating quantitative research findings. International Studies Review, 11, 707-735.

Englebert, P. (2000). Solving the mystery of the Africa dummy. World Development, 28(10), 1821-1835.

Fearon, J. D. (2006). Ethnic mobilization and ethnic violence. In B. R. Weingast, & D. Wittman (Eds.), Oxford handbook ofpolitical economy (pp. 290-332). Oxford: Oxford University Press.

Fearon, J. D., & Laitin, D. D. (2003). Ethnicity, insurgency, and civil war. American Political Science Review, 97, 75-90.

Gandhi, J., & Lust-Okar, E. (2009). Elections under authoritarianism. Annual Review of Political Science, 12, 403-422.

Ghobarah, H. A., Huth, P., & Russett, B. (2004). The post-war public health effects of civil conflict. Social Science & Medicine, 59(5), 869-888.

Giroux, J., Lanz, D., & Sguaitamatti, D. (2009). The tormented triangle: The regionalisation of conflict in Sudan, Chad, and the Central African Republic. In Crisis States Research Centre workingpapers series 2 (Vol. 47). London, UK: Crisis States Research Centre, London School of Economics and Political Science.

Gleditsch, N. P., Wallensteen, P., Eriksson, M., Sollenberg, M., & Strand, H. (2002). Armed conflict 1946-2001: a new dataset. Journal of Peace Research, 39(5), 615-637.

Green, E. (2012). On the size and shape of African states. International Studies Quarterly, 56(2), 229-244.

Gubler, J., & Selway, J. (2012). Horizontal inequality, crosscutting cleavages, and civil war. Journal of Conflict Resolution, 56(2), 206-232.

Gupta, A., & Ferguson, J. (1992). Beyond "culture": space, identity, and the politics of difference. Cultural Anthropology, 7(1), 6-23.

Gurr, T. (1993). Why minorities rebel: a global analysis of communal mobilization and conflict since 1945. International Political Science Review, 14(2), 161-201.

Gurr, T. (2000). Peoples versus states: Minorities at risk in the New Century. Washington, DC: United States Institute of Peace.

Hegre, H., Ellingsen, T., Gates, S., & Gleditsch, N. P. (2001). Towards a democratic civil peace? American Political Science Review, 95(1), 33-48.

Hendrix, C. (2010). Measuring state capacity: theoretical and empirical implications for the study of civil conflict. Journal of Peace Research, 47(3), 273-285.

Herbst, J. (2000). States and power in Africa: Comparative lessons in authority and control. Princeton: Princeton University Press.

Kaldor, M. (1999). New and Old Wars: Organized violence in a global era. Cambridge: Polity Press.

Kalyvas, S., & Balcells, L. (2010). International system and technologies of rebellion: how the end of the Cold War shaped internal conflict. American Political Science Review, 104(3), 415-429.

Kalyvas, S., & Kocher, M. (2007). How "free" is free riding in civil wars? Violence, insurgency and the collective action problem. World Politics, 59(2), 177-216.

Kasara, K. (2007). Tax me if you can: ethnic geography, democracy, and the taxation of agriculture in Africa. American Political Science Review, 101(1), 159-172.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (1999). The quality of government. Journal of Law, Economics and Organization, 15(1), 222-279.

Le Billon, P. (2001). The political ecology of war: natural resources and armed conflicts. Political Geography, 20(5), 561 -584.

Lemarchand, R. (2009). The dynamics of violence in Central Africa. Philadelphia: University of Pennsylvania Press.

Lijphart, A. (1977). Democracy in plural societies. New Haven: Yale University Press.

Mozaffer, S., Scarritt, J., & Galaich, G. (2003). Electoral institutions, ethnopolitical cleavages and party systems in Africa's emerging democracies. American Political Science Review, 97(3), 379-390.

Murshed, M., & Gates, S. (2005). Spatial-Horizontal inequality and the Maoist insurgency in Nepal. Review of Development Economics, 9(1), 121-134.

0stby, G. (2008a). Polarization, horizontal inequalities and violent civil conflict. Journal of Conflict Resolution, 45(1), 143-162.

0stby, G. (2008b). Inequalities, the political environment and civil conflict: evidence from 55 developing countries. In F. Stewart (Ed.), Horizontal inequalities and conflict: Understanding group violence in multiethnic settings (pp. 136-159). Basingstoke: Palgrave Macmillan.

0stby, G., Nordas, R., & Rod, J. K. (2009). Regional inequalities and civil conflict in Sub-Saharan Africa. International Studies Quarterly, 53(2), 301 -324.

0stby, G., & de Soysa, I. (2008). Too weak to fight? Horizontal inequality and state reperssion, 1980 - 2004. Unpublished Manuscript. Oslo, International Peace Research Institute.

Posner, D. (2004). Measuring ethnic fractionalization in Africa. American Journal of Political Science, 48(4), 849-863.

Raleigh, C. (2010a). Political marginalization, climate change, and conflict in African Sahel states. International Studies Review, 12(1), 69-86.

Raleigh, C. (2010b). Seeing the forest for the trees: does physical geography affect a state's conflict risk? International Interactions, 36(4), 384-410.

Raleigh, C. (2014). Pragmatic and promiscuous: political militias across African states. Journal of Conflict Resolution (forthcoming).

Raleigh, C., & Hegre, H. (2009). Population size and civil war. A geographically disaggregated analysis. Political Geography, 28(2), 224-238.

Raleigh, C., & Kniveton, D. ((1)2012). Come rain or shine: an analysis of conflict and climate variability in East Africa. Journal of Peace Research, 49(1), 51-64.

Raleigh, C., Linke, A., Hegre, H., & Carlsen, J. (2010). Introducing ACLED: an armed conflict location and event dataset. Journal of Peace Research, 47(1), 651-660.

Reno, W. (1998). Warlord politics and African states. Boulder, CO: Lynne Rienner.

Reno, W. (2011). Warfare in independent Africa. Cambridge: Cambridge University Press.

Sambanis, N. (2004). What is civil war? Conceptual and empirical complexities of an operational definition. Journal of Conflict Resolution, 48(6), 814-858.

Scarritt, J. R., & McMillan, S. (1995). Protest and rebellion in Africa: explaining conflicts between ethnic minorities and the state in the 1980s. Comparative Political Studies, 28, 323-349.

Scarritt, J., & Mozaffar, S. (1999). The specification of ethnic cleavages and ethno-political groups for the analysis of democratic competition in contemporary Africa. Nationalism and Ethnic Politics, 5(1), 82-117.

Schedler, A. (2006). Electoral authoritarianism: The dynamics of unfree competition. Boulder: Lynne Reinner.

Sobek, D. (2010). Good for the money: international finance, state capacity, and internal armed conflict. Journal of Peace Research, 49(4), 391 -405.

Stewart, F. (2008). Horizontal inequalities and conflict: an introduction and some hypotheses. In Stewart (Ed.), Horizontal inequalities and conflict: Understanding group violence in multiethnic settings (pp. 1-20). Basingstoke: Palgrave Macmillan.

Stewart, F., Brown, G., & Langer, L. (2008). Major findings and conclusions on the relationship between horizontal inequalities and conflict. In Stewart (Ed.),

Horizontal inequalities and conflict: Understanding group violence in multiethnic settings (pp. 301-325). Basingstoke: Palgrave Macmillan.

Straus, S. (2011). Wars do end! Changing patterns of political violence in Sub-Saharan Africa. African Affairs, 111(443), 179-201.

Thies, C. G. (2009). National design and state building in sub-Saharan Africa. World Politics, 61(4), 623-669.

Thies, C. (2010). Of rulers, rebels, and revenue: state capacity, civil war onset, and primary commodities. Journal of Peace Research, 47(3), 321-332.

Tilly, C. (1985). War making and state making as organized crime. In P. Evans, D. Rueschemeyer, & T. Skocpol (Eds.), Bringing the state back in (pp. 169-191). Cambridge: Cambridge University Press.

Treisman, D. (2007). The architecture of government: Rethinking political decentralization. Cambridge, UK: Cambridge University Press.

Weinstein, J. (2007). Inside rebellion: The politics of insurgent violence. Cambridge: Cambridge University Press.

Wilson, S., & Butler, D. (2007). A lot more to do: the sensitivity of time-series cross-section analyses to simple alternative specifications. Political Analysis, 15(1), 101-123.

Wimmer, A., Cederman, L. E., & Min, B. (2009). Ethnic politics and armed conflict: a configurational analysis of a new global dataset. American Sociological Review, 74(2), 316-337.

World Bank. (2009). World development indicators. Washington D.C: World Bank.

Young, C. (1994). The African colonial state in comparative perspective. New Haven, CO: Yale University Press.