The Last Drop: how Water Scarcity Leads to Communal Conflict in the Horn of Africa

(1)

The Last Drop: how Water Scarcity Leads to

Communal Conflict in the Horn of Africa

MASTER THESIS Crisis and Security Management

Author : Mariese Broekhuisen

Student ID : s1792202

Supervisor : Dr.mr. E.T. Aloyo

Second reader : Dr.mr. E.E.A. Dijxhoorn

(2)

The Last Drop: how Water Scarcity Leads to

Communal Conflict in the Horn of Africa

Mariese Broekhuisen

June 7, 2020

Acknowledgements

The completion of this thesis could not have been possible without the help and assistance of a few people, whom I would like to thank.

I am profoundly grateful to Charlotte Grech-Madin, Ph.D., and Mr. Stefan Döring, both from Uppsala University, for their support for my research, their recommendations, and their response to my inquiries.

Further gratitude goes out to my supervisor, Dr.mr. E.T. Aloyo, for his support and feedback during the thesis process and for his time. On the same note, I would like to thank second reader Dr.mr. E.E.A. Dijxhoorn for the time and effort he will invest in his

assessment of my work. Your contributions are very much appreciated and met with my sincerest gratitude.

I am very much thankful to my family, who have provided me in many ways with the emotional support and assistance that was at times necessary to get through the process.

Finally, I would like to thank Mr. Jasper Krause for his continuous support,

reassurance, advice, hearing my many questions, and taking the time to proofread my work. I am extremely grateful.

(3)

Abstract

This thesis uses causal process tracing (CPT) to analyse the relationship between water scarcity and incidences of communal violence. As previous research has mainly focused on rainfall as a factor determining water scarcity and the intrastate level of conflict, this research builds upon georeferenced, intrastate data of communal conflict in the Horn of Africa and georeferenced data of precipitation, groundwater access, temperature, drought and surface water presence as determinants of water scarcity and the hydrological cycle. Within this region, many livelihoods are heavily dependent on such water access. Furthermore, the low level of development leaves peoples more vulnerable and exposed in this region. Water scarcity is hypothesized to increase incidences of communal violence. Based on the types of evidence collected for CPT, it is hypothesized that water scarcity leads to Internally

Displaced Persons (IDPs), which contributes to such an increase in communal conflict incidences. The effect is also argued to be enhanced by a lack of state presence, high

population density, and highly water-dependent livelihoods. These propositions are analysed and processed to form a hypothesized causal configuration. The result shows that a lack of water access is indeed associated with a higher risk of communal violence, but that this effect is conditioned by water dependency, a lack of coping capacity, and high vulnerability as a result of a lack of state presence. Resultingly, this research suggests that IDPs and population density cause tensions to rise due to high water stress levels and the competition over this research, which is found to be the main driver in the relation between water scarcity and communal conflict.

(4)

Chapter 1: Introduction

There has been a great increase in studies trying to establish possible relationships between climate change and armed conflict. As such, changes in temperature and

precipitation are expected more frequently due to climate change and both researchers and national security and policy circles have started to accept it as a threat to peace and security. Extreme events such as droughts could have severe implications for society, which may leave great numbers of people exposed to water stress (Detges, 2016; Von Uexkull, Croicu, Fjelde & Buhaug, 2016).

Despite the amount of studies, however, a consensus has not yet been reached on the connection between climate change and violent conflict (Von Uexkull, Croicu, Fjelde & Buhaug, 2016). When it comes to Africa, its dependence on precipitation, its social divisions, and its weak coping abilities should make it especially vulnerable to the changes in weather and conflicts. Findings suggest, however, that evidence of a relationship between climate change and conflict in Africa is generally quite weak (Salehyan, 2014).

After the seasonal rains of late 2018 and 2019 failed to bring much moisture to East Africa, a lot of droughts can be found in the region. Such a lack of precipitation has contributed massively to a food security crisis. Many people in East Africa lack sufficient food from time to time. These rainfall deficits eventually also result in soil moisture deficits. This means that rainfed crops, specifically during planting times and growing seasons, can be devastated as a result of moisture deficits. NASA’s satellite data and land surface models inform drought monitoring systems, which show that the Horn of Africa has below-normal soil moisture. Its dryness was caused by the rainfall being below average during its longest of two annual periods of rain, the ‘Gu’ (Hansen, 2019). The Famine Early Warning System Network (FEWS NET) has targeted the region for its dryness, claiming the Gu rainfall was about half of its usual amount through May 2019. The short rainy period called ‘Deyr’ that preceded this Gu season also only generated below-average precipitation from October until November. Only a few measuring stations in the north of Somalia recorded mere significant rainfall accumulations in April, after which the UN Food and Agriculture Organization warned 2 million people in Somalia were facing severe hunger. The country has faced long periods of drought in the past decades, but FEWS NET now reports it might reach an ‘emergency’ phase for food security, which is one phase away from the stage of ‘famine’. Ethiopia and Kenya are expected to reach the prior stage of ‘crisis’ (Hansen, 2019).

(6)

Inconsistency in existing research has originated because, amongst other reasons, local vulnerability and coping abilities that determine the effect drought has on the community have not sufficiently been taken account. Especially long periods of sustained drought, as has been the case for the Horn of Africa (HOA), can severely undermine alternative coping mechanisms. What makes individuals especially vulnerable to droughts is their reliance on agriculture fed by rain which generates their income and provides them with food. Areas that experience sustained periods drought or depend heavily on precipitation for agricultural purposes are more likely to see conflict as a result of drought (von Uexkull, 2014). As a result of such dependence on water, irrigation schemes have been developed. Irrigation dams, canals, and boreholes have been used to secure water during the year, which means more water diversion from wetlands or more groundwater extraction (UNEP, 2010). Therefore, to find out the societal consequences of water scarcity, it is very important to not only look at precipitation but to also look at more reliable water sources, such as groundwater (Döring, 2020).

The HOA is one of the most disputed of the regions of the globe, as it has experience with great numbers of conflicts varying in kind and knows great instability (Mekonnen Mengistu, 2015). It has particularly seen an increasing number of communal conflicts over the past 15 years, and the agricultural sector plays a great role in Ethiopia, Kenya, and

Somalia (van Weezel, 2016). This paper will, therefore, research how drought affects the rate of communal violence in the Horn of Africa. The research question therefore reads:

How has water scarcity affected the rate of communal conflictin the Horn of Africa between 1990 and 2014?

Every year there are several incidences of armed nonstate conflict that occur without direct state involvement. In complex civil wars, such as the current one in Somalia which started in 1991, clashes between rebel forces are very important determinants of armed nonstate conflict. In Kenya, however, civil conflicts against the state are ways less common, yet nonstate conflicts between communal militias or gangs have devastating effects on the country (von Uexkull & Pettersson, 2018). Taking into account all sources of water that could be involved in the mitigation of dry spells, the research will include groundwater, precipitation, ground moisture, and temperature, as well as major water body presence (Döring, 2020). The research will compare Somalia, Kenya, and Ethiopia for they are the countries with most communal conflict incidences (von Uexkull & Pettersson, 2018),

(7)

granting a wider spread of locations and issue origins to compare to get a more reliable result. Seeing as causal process tracing (CPT) will be used to answer the research question, this subsequently allows us to be able to have an accurate hypothesized causal conjunction that is based on multiple conflicts of different natures. Furthermore, Ethiopia, Kenya, and Somalia are all countries whose agricultural sectors are very important, with especially many

agropastoralist communities relying on water to get by as they both farm crops and rely on livestock for produce (Döring, 2020). The selected period delineates the recent developments of drought and violence within the region, covering the famine and food crisis that hit

Somalia from 2010 to 2012 (BBC, 2013), and covers the most actual available data. It also allows us to give enough of a historical and temporal account of developments in the region, which is necessary for causal process tracing (Blatter & Haverland, 2012). By using different sources of data from different places and people, data triangulation is achieved for the

collecting of fuller, richer data (Wilson, 2014). This ensures the data is reliable and not based on extrapolation or other estimates, but instead based on accurate measurements and

statistics. By employing CPT and based on this case selection, this research will contribute to an emerging field linking water scarcity to violent conflict by non-state actors and add the Horn of Africa to previous analyses. Expected is that drought because difficulty in accessing groundwater, low precipitation, and lack of major river presence does make communal violence more likely. The results of this study could contribute to lower conflict rates by establishing the link between drought and communal conflict, which might form a preventive basis.

Chapter 2: Theoretical framework

Water Scarcity

When it comes to explaining communal conflict, resource scarcity is about

relationships between resource availability and population growth. From a neo-Malthusian standpoint, this means that when resources become scarce, or when they have been depleted and overused up until a certain level, this promotes conflict. Environmental changes and the resulting scarcity of natural resources is decisive in the rise of conflicts (Hagmann, 2005).

The general definition of drought is a deficit of water that appears, a temporal anomaly when compared to long term conditions. Droughts in Africa have been known to cause

(8)

problems such as, but not limited to, failed crop harvests, food shortages, famine, epidemics, and mass migration. The phenomenon of drought and its complexities affects both natural environments and socioeconomic systems. Recently, severe events have occurred globally. In 2011, for example, there was the Horn of Africa Drought, and in in 2014 there was a drought in North China. After many of these droughts, negative effects on natural and socioeconomic systems were reported (Peng et al., 2019).

Generally, droughts take the form of precipitation deficiency, soil moisture deficiency, groundwater deficiency, a socioeconomic response to water supply and demand, and

environmental or ecologic effects (Peng et al., 2019). Rainfall strongly influences water availability and other resources, such as crops and livestock. Rural communities have adapted and learned to account for the fluctuations across seasons and years and plan their crops and herd movements around it. Several ways to cope with dry periods have been devised, however, rainfall reaching far below expectations might eliminate such strategies and result in a loss of incomes, food insecurity, migration, and poverty (Detges, 2016).

Under-development harms the adaptation capacities of people living in a region, which often leads to their displacement and more instability within a region during times when water is scarce (IMDC, 2017). Rainfall is mainly used for agriculture, yet groundwater serves additional purposes, such as providing a source of water for small-scale family farms that make up around 70 percent of the agricultural land in Africa. Rural communities are highly dependent on groundwater in running their households, and it is essential for livestock and drinking water. Irrigation on the basis of groundwater has also become more common in recent years, as all countries have been provided with groundwater irrigation systems. Refugee camps often have groundwater as their only water source, which makes them very vulnerable to shortages and contaminations (Döring, 2020).

Despite major disasters having a strong visible effect on impacted communities, it is small, recurring effects that have proven particularly harmful to development and growth in countries with middle to low income levels according to Ban Ki-moon, former United Nations (UN) secretary-general (IDMC, 2017). Such effects include rainfall patterns, which has been a major cause of displacement in Africa (IDMC, 2017). Ethiopia in particular has seen a steady rise in its displacement figures, the drivers of which have historically been said to be a combination of disasters both with a slow and sudden onset, resource competition, and ethnic tensions (IDMC, 2017). The Horn of Africa is an area suffering from droughts, often seeing consecutive years of it (IDMC, 2017). It has led to great levels of resource competition in the region, in particular between pastoralists and farmers. This gets reinforced by the

(9)

ethnic tensions that have long been occurring both within and across its borders (IDMC, 2019). The livelihoods of pastoralists and farmers have been severely undermined by the loss of pastoral lands livestock deaths and low crop yields resulting from scarce water. This has resulted in a serious problem regarding displacement and has also led to great food insecurity (IDMC, 2019). Particularly in Ethiopia’s Somali region and going into Somalia water

shortages and resulting food shortages form a major issue. Drought has caused heightened vulnerability (IDMC, 2019).

Displacement as a result of drought is localized, as people tend to move within their own region. Refuge is sought after in neighbouring communities whilst other IDPs choose to find shelter in camps scattered across the region (IDMC, 2017). The Horn of Africa is one of the regions most affected by drought on the continent of Africa (IDMC, 2019), and

displacement in this region is characterised by the inability and vulnerability of populations impairing their coping mechanisms when their livelihoods are devastated by droughts. This goes hand in hand with a strongly increased resource competition, which often leads to further displacement associated with conflict (IDMC, 2017).

Communal conflict

It has been commonly acknowledged that the issue over which involved parties fight shapes the causes, dynamics, and consequences of armed conflict. When it comes to

territorial issues, for instance, it might be challenging to form a resolution. In the same way, characteristics of the involved actors, such as assistance received from other parties, greatly affect dynamics and the risk of recurrence of conflict. When it comes to nonstate conflicts there are three main categories of conflicts that occur: conflicts over authority, conflicts over territory, and other issues (von Uexkull & Pettersson, 2018). Nonstate conflicts over authority could for instance be conflicts between political candidate supporters, or conflicts between supporters of different leaders within a group or community such as religious groups or rebel groups. The most frequent conflict issue, however, is territory (von Uexkull & Pettersson, 2018). A great number of land-use conflicts have been observed where water or agricultural lands are the source of disagreement. Oftentimes agricultural land or water resources are an issue at stake, such as happened when the Borana and Turkana pastoralist groups fought each other in 2011 after the latter accused the former of taking away their grazing land. Other issues within this category concern territorial issues such as administrative borders such as clashes over homelands or different ethnics groups’ border domination (von Uexkull &

(10)

Pettersson, 2018). Within the category of ‘other issues’ we also find livestock issues and livestock raiding with resulting deadly violence, which is a frequent phenomenon in the Sahel region with its pastoralist cultural roots. Kenya in particular has the highest number of

conflicts of this nature. Religious issues in nonstate conflicts appear to be rare and appear to often be combined with other reasons such as territory. Such was the case between Somali groups Ahlu Sunna Waljamaca and Al-Shabaab between 2008 and 2010 (von Uexkull & Pettersson, 2018).

In Somalia and Ethiopia, livestock farmers make use of different well types (Bogale & Korf, 2007). Such findings can be linked directly to communal conflict because it sheds a light on resource-based conflicts, specifically relevant for the link between water and conflict and its relation to competition over land use (Döring, 2020). In Africa, such land use

questions can be rather vague considering the nature of its ownership types which know great variance even within small villages. This can lead to severe friction between the various users of land (Turner & Moumouni, 2018). However, farmers and herders often choose a more peaceful approach and opt for informal conflict resolution to prevent an escalation of violence (Bogale & Korf, 2007). Nonetheless, disputes do turn violent and specifically tend to when there is a certain negative influence on the mechanisms behind informal conflict resolution, of which biased local politics has found to be one. In the Horn of Africa, farmer-herder conflicts have seen dynamics between shifts in local politics and different livelihoods, which gives more insight into the motivation for use of violence by non-state actors (Döring, 2020).

Even though the majority of communal violence results from disputes over land that are caused by a lack of or inaccurate state control and unclear systems regarding land-rights, the conditions brought about by droughts, for example, create favourable conditions for the recruitment of people that have lost their livelihoods by rebel forces. During the Somalian famine that started in 2010, a number of Somalian pastoralists joined an extremist

organization. However, drought can also lead to clashes between communal groups. Water scarcity restraints essential resources, which increases competition. Such tensions could arise in the instance of herders forced to stay in areas that are used for agriculture during severe dry spells. This could also result in predatory behaviour, as has happened in the drylands of Kenya, where cattle raiding is used to compensate for economic losses amongst herders (Detges, 2016). Such cattle raiding is also a form of power display and pride to some of the pastoralist groups, and may sometimes even include murder (Döring, 2020). Drought may

(11)

also force people to migrate or travel greater distances to be able to obtain their resources, making them more susceptible to attacks (Degtes, 2016).

Armed conflict in relation to water scarcity mainly concerns clashes between

communal groups, without the state being involved. The mobilization of such actors is often based on religious affiliations or ethnicity, relationship ties, or livelihood. A lot of disputes over land, for example, arise between farmers, who rely on their land to grow crops, and pastoralists, who need the land to let their livestock graze on it. Drought has the ability to destroy harvests and even turn arable land into desert, which lowers the agricultural

production levels. The resources available for both groups to live and accumulate income can thus be heavily impacted by droughts. Two groups with the same livelihoods can also clash in competition for, for example, arable land, which is the main source of conflict between pastoral groups in the Horn of Africa (Fjelde & von Uexkull, 2012). Access to water is the most important driver of relations between farmers and herders in Africa. Across Africa land control is a difficult subject, as even small villages with land tied to them know various types of ownership, which can create quite some friction between different types of land users. Such disputes, even when most are peaceful, can turn violent. Research has shown that non-state conflict has mainly involved disputes over territory, authority, and resources that can be looted. When it comes to water, research in Northern Kenya has shown that fighting is more likely close to well sites. Extreme precipitation, both high and low, has also been found to increase the risk of communal conflict (Döring, 2020).

Furthermore, the economic vulnerability of communities goes hand in hand with physical vulnerability in the region. Poor communities often retreat to land that is

marginalized and degraded and find themselves more exposed to environmental degradation (Fjelde & von Uexkull, 2012). The choice of using violence when faced with hardships induced by environmental factors is thus highly dependent on poverty. Additionally, poverty is one of the main predictors of communal violence as a result of the sentiments of

deprivation, frustration, and aggression such communities may feel. Such sentiments are good facilitators for the use of violence against other communities (Fjelde & von Uexkull, 2012). The sacrifice such poor communities make when engaging in violence is also lower compared to communities that are better off, seeing as the income they lose when they choose to engage in violence is very limited (Fjelde & von Uexkull, 2012).

(12)

Linking Water Scarcity to Communal Conflict

Studying the relationship between water scarcity and communal conflict has started to enter the research field increasingly over the past years (Döring, 2020). Enhanced data access has opened up a space for analysis on the sub-national level. Many studies have focused on the effects of climate change or water in combination with other aspects of climate

variability, yet water has not played a major role on its own so far (Döring, 2020). However, a study on pastoralist violence by Detges (2014) has found fighting to be more likely in locations in Northern Kenya that are close to well sites. Other studies, however, find communal conflict more likely in areas with anomalous precipitation patterns. When combining such precipitation patterns with conflict in parts of East Africa, several studies have found results that are inconclusive (Döring, 2020). Fjelde & von Uexkull (2012) found deviations from precipitation patterns increase the risk of communal conflict if they deviate too much with unfavourable effects. Such deviations could be unusually long wet periods as well as dry periods. Nonetheless, when assessing water conflicts, rainfall, and temperature have not been found to explain fully why the disputes arose (Döring, 2020).

However, there does seem to be a pattern of agreement that other factors are determinant when it comes to the effect of water scarcity on communal violence. For example, it seems to be more likely for violence to escalate following drought in situations where the population already lacks access to water infrastructure, or amongst groups that are politically marginalized. Non-state conflict also seems more likely in countries with high levels of equality, or countries that are on the edge of a regime change (Döring, 2020). A direct relationship between droughts and communal violence has not been described unless met by favourable conditions such as weak institutions and a government that does not respond. Other contextual factors to determine farmers’ and herders’ susceptibility to drought are, for example, their ability to adapt and access services that are essential to them. Losses caused by drought will contribute to violence depending on the role of the state and civil society actors in preventing disputes (Detges, 2016). Whether scarcity can harbour violent conflicts and lead to violence thus depends on the socio-economic context of a region (Hagmann, 2005). The pre-existing vulnerability of livelihoods, including their capacity to anticipate environmental hazards, and manage and recover from them, therefore, plays an important role in determining the effect of drought. What plays the largest role is the

dependence on water for food and income. Another determinant is the longevity and severity of such exposure (von Uexkull, 2014).

(13)

As water scarcity refers to a physical lack of water resources for a certain geographic area and its inhabitants, such scarcity can be explained by several difficulties that may arise in the water sharing process. Such scarcities may be due to climatic processes, such as drought spells, but it can also be due to financial constraints when it comes to transport on infrastructure costs. The exact links between water scarcity and violence have not been grasped fully yet, however, it is important to look into weakened state presence, weak adaptation capabilities and local migration in the areas to see assess what, if any, influence they have in the occurrence of communal conflict (Döring, 2020). Water scarcity does not only worsen farming, herding, and agricultural conditions, it also imposes high levels of water stress on people’s living conditions and limits access to nutrition and sanitation, compromising health. In circumstances where state involvement may create tensions, or where the government is delegitimized where it is not involved, resettling to different

surroundings by means of migration might pose a feasible solution. However, this eventually leads to suffering, where friction might also arise (Döring, 2020). Livestock husbandry, or pastoralism, for example, has been neglected when it comes to governance (Turner et al., 2016). Such mobile groups are often restrained by local authorities’ reluctance to meet its seasonal needs, which causes a lot of movement across several territories. Informal and formal governance systems are known to have a lot of failures resulting in the

mismanagement of resources like water points and pastures, even if such systems are meant to mediate seasonal access to those resources that are needed in agropastoralist areas (Turner et al., 2016). One of the reasons behind this is the mismatch between mediation forums that are part of local jurisdictions and the large geographical scales at which livestock need to move to be able to access food (Turner et al., 2016). In regions where there is a severe lack of available water, especially due to a lack of rainfall, livestock simply has to move to be able to access food (Turner et al., 2016).

State presence is dependent on the risk reduction mechanisms, governance,

communication, physical infrastructure, and other developmental factors (Martin-Ferrer, Vernaccini, & Poljansek, 2017). These factors are all important when it comes to assessing the coping capacity of a population. Furthermore, this is highly related to socioeconomic (in)equality and the vulnerability of groups (Martin-Ferrer, Vernaccini, & Poljansek, 2017). This means that the risk of communal conflict as a result of water scarcity is mitigated by the amount of state presence in a certain area. If people have a higher capacity to cope with water scarcity due to measures and systems installed by the state, they will be affected less than if they were to not have any of those measures in place (IDMC, 2017). This means that

(14)

underdeveloped areas have a higher risk of communal conflict as a result of water scarcity (Martin-Ferrer, Vernaccini, & Poljansek, 2017). Others argue that governance and

institutionalized, formal rules have no positive effect on the effects drought could have in relation to violence. Linke et al. (2015) argue that the idea that drought leads to an increase in support of communal conflict and violence is moderated or ameliorated by state presence is incorrect. In fact, based on surveys they conducted with members of several communities in Kenya, they find that in times where drought is reported to be increasing negatively, thus getting worse, they find that communication within the community is more helpful in lowering levels of support for violence (Linke et al., 2015). This means that where inter-communal communication is strong, the effects of drought and the risk of inter-communal conflict occurrence are ameliorated (Linke et al., 2015). However, this means that such communities do need to have certain communication mechanisms in place, and other mechanisms to help them cope with the negative effects of drought. In underdeveloped areas this may be very difficult. Linking violence back to poverty and relating it to water scarcity, poor communities find themselves more incapable of withstanding such environmental change (Fjelde & von Uexkull, 2012). As poverty narrows the resources and options for those communities to be able to cope with the economic consequences of water scarcity. In combination with the limited assets, land ownership uncertainty, often single income reliance, and market access restriction, this leaves poor communities very vulnerable to environmental hardships, including water scarcity (Fjelde & von Uexkull, 2012). Coping mechanisms often rely on state presence and installed mechanisms. Poverty thus imposes a great restriction on the range of options poor groups have at their disposal when dealing with water scarcity, which means they are at higher odds of turning towards violence (Fjelde & von Uexkull, 2012). This relates to communities’ political relevance and its determination in water scarcity resilience. It is characteristic of African politics that groups are formed along ethnical lines when it comes to economic and political demands. Distribution of resources, therefore, tends to happen along those lines. This means groups are easily excluded from power and

resources, which can be seen as another strong cause of communal violence (Fjelde & von Uexkull, 2012). Ethno-political groups that lack representation in the political arena experience sentiments and grievances that are likely to lead to violent participation. Such excluded groups also only have limited response availability (Fjelde & von Uexkull, 2012). Political exclusion may lead to violent communal conflict as a result of water scarcity in the following ways: first, as state resources are distributed based on groups’ political relevance, excluded groups often reside in areas with weak state presence in which there is a lack of

(15)

administration which could have a conflict resolution function, public services, physical infrastructure (Raleigh, 2010). Second, a group’s political status highly influences the response the government has in the event that water scarcity occurs. Groups that have established political relevance are more likely to receive help in terms of disaster relief, infrastructural investments, and policies aimed at economic compensations. However, when excluded groups are faced with hardships such as water scarcity, they receive little to no help on the basis that they do not directly threaten the regime’s political stability (Raleigh, 2010). Exclusion thus reinforces the vulnerability of communities. This creates more susceptibility to poverty and migration, and will often result in such groups resorting to violent strategies as a result of water scarcity or other environmental hardships (Raleigh, 2010). Violence in these areas is oftentimes seen as a way for these communities to govern themselves, and conflicts over access to resources is a product of this. Where state presence is minimal, communal conflict and the use of violence is done to mediate resource access, which is necessary to be able to sustain the livelihoods for these groups (Raleigh, 2010). This thus means that a lack of state presence will make it more likely for politically excluded groups to resort to violence considering their range of options is otherwise very narrow. Marginalization and exclusion both politically and economically thus will increase communities’ exposure to the

consequences of water scarcity, such as droughts and decreased precipitation. Whereas states usually tend to offer coping strategies through economic policies, in the case of such

excluded groups such offers are limited (Fjelde & von Uexkull, 2012). In this sense, water access competition might also intensify when it comes to local communities, especially when they have been excluded politically. They find themselves in a situation where the

opportunity cost of using violence is not high enough to hold them back. Especially in combination with the heightened sentiments of frustration and resentment, specifically against groups that are better off, there is a higher chance of communal violence occurring (Fjelde & von Uexkull, 2012). High poverty levels thus enhance communities’ vulnerability and thus contribute to the lack of coping capacity. This makes it more likely that such communities will resort to violence to gain access to water resources or to resolve high competition tensions. Group exclusion enhances this likeliness to resort to violence even further.

However, not only poor and excluded groups suffer the consequences of water scarcity. Reduced water access worsens conditions for farmers, herders, and those who rely on

agricultural produce; nutrition and sanitation rely on water and without it, living conditions in general would be impaired. It is clear, however, that not all areas that are scarce in water

(16)

witness communal conflicts. Several institutions are in place in several regions that might mitigate disputes relating to water. This means that state presence should decrease the likelihood of such communal violence occurring, and vice versa. Contrary to this finding, however, the state’s actions might often fail to reach the entire population, therefore leaving water to be a local issue (Döring, 2020). This again enhances vulnerability amongst local marginalized groups.

Most research trying to establish a connection between any kind of environmental source and violence by means of communal conflict does not find a direct relationship, however, does state the importance of structures in place to mediate the relationship between communities and such sources (Fjelde & von Uexkull, 2012). To be able to find a link, therefore, local wealth configurations and political influence need to be analysed to be able to understand the vulnerability of a community (Fjelde & von Uexkull, 2012). Research

suggests that water scarcity is likely to increase communal conflict risks in areas where populations are more dependent on the weather, and thus have more to lose compared to periods in which the weather is favourable for their livelihood (Bell & Keys, 2018). It can be concluded from this that communities whose herds or crops are heavily dependent on

favourable weather are more heavily impacted than communities in which this does not play as significant a role.

Adaptation processes as a reaction to shortage by finding ways to become less water-reliant could inhibit the link between water scarcity and violent conflict. However, such adaptation strategies have no infinite success rate. Such has shown in Kenya where there have been various conflicts over stream diversion and grazing rights over the past century. To some groups using violent means seems to be a feasible adaptation strategy to appropriate water where groundwater is often overexploited, and various recurring droughts occur (Döring, 2020). A phenomenon that illustrates the limitations of such adaptation strategies is local migration. Indeed, this does not mean that violence might still occur in case of scarcity. Such migration often takes place within administrative boundaries of a smaller scale, even local, especially when it is a temporary means to cope with droughts. It is always preferred to move somewhere nearby for short-term adaptation. Water scarcity may thus result in

Internally Displaced People of all sorts, which may lead to disputes as a result of changed societal structures by new interactions, land redistributions, or changes in power distributions amongst ethnic groups, which could also result in violent communal conflict (Döring, 2020). Rural areas often use rainfall, surface water, and groundwater combined. Especially the latter is vital to battle drought to provide the cattle or in goat herding. The Turkana groups in

(17)

Kenya consider water availability when raiding, specifically also looking into whether other ethnic groups may have used water points that already existed. Pastoralists in Ethiopia indicate water source claims by slaughtering animals close to boreholes (McCabe, 2004; In Döring, 2020).

The IDMC (2016) developed a simulator for pastoralist livelihood and displacement to explore the displacement of pastoralist groups associated with drought. This simulator is based on the hazard that drought brings and the vulnerability of the people, modelling a causal chain between low rainfall levels and displacement. Summarized, it concludes that rainfall in combination with causal conditions land access and pasture upgrading are positively related to livestock which is a source of income. Income, correlated with

remittances, livestock markets, and cash assistance, in turn, is positively related to livestock, meaning higher income is more livestock. This means that with high levels of rainfall, pastoralists prosper, whereas with low levels of rainfall the pastoralist practices collapse, and this results in drought IDPs (IDMC, 2016). Displacement will likely increase as populations are moving to obtain water and livelihoods. Via this way, water scarcity may thus also be a driver of conflict (IDMC, 2017). Ethiopia in particular has seen long periods of droughts resulting in the displacement of many pastoralists and agropastoralists, and the HOA has seen repeated communal conflicts that have contributed even further to IDP levels (IDMC, 2019). In addition to an increase in the level of IDPs, increased competition over water as a resource and loss of livelihoods resulting in disenfranchising may also lead to conflict (IDMC, 2017). Research that has so far been done concerning IDPs has shown that rainfall shocks indeed lead to circumstances in which conflict is more likely (IDMC, 2017).

The presence of the state in forming resolutions and its relation to the coping capacity of communities and short-range migration in the form of internal displacement thus seem to form two causal pathways that may help explain the link between water scarcity and

communal violence (Döring, 2020). Existing literature and research have covered farmer-herder conflicts, however, there has so far not been an in-depth analysis of water scarcity that includes other factors beyond drought for the Horn of Africa. Furthermore, none of these studies have used georeferenced data to analyse the subnational level. Since there is only limited data, a first step is starting by uncovering the available data concerning water scarcity and including groundwater that is used based on its more widespread availability (Döring, 2020). This research, therefore, aims to analyse the role of the various conditions that determine water scarcity, including groundwater, surface water access, drought and

(18)

and violent communal conflict, such as state presence and its role in enhancing population’s coping capacity, Internally Displaced Persons, but also population density (Döring, 2020). Water scarcity as a result of different types of shortages will be compared to incidences of communal conflict. As water scarcity may cause people to move to other nearby places for the short-term, this can cause conflict elsewhere due to power shifts or new interactions from causing IDPs as a result (Döring, 2020). Such situations are prone to result in violence. The following hypotheses thus are proposed:

H(I). Water scarcity in an area increases the incidence of communal conflict.

H(Ia). Water scarcity leads to Internally Displaced Persons (IDPs) which increases the

incidence of communal conflict.

Based on the theory, state presence is very influential on the effect of water scarcity on an area as it affects how water is managed but also divided. In Somalia, for example, the civil war ending meant the collapse of the government after which clans and ethnic groups began putting taxes on water resources, making many herders pay large sums for land farmers had claimed as theirs. This led to a lot of violent incidences (Döring, 2020). Furthermore, in 2009 over 60,000 people were displaced in Southern Ethiopia over disputes concerning borehole construction (BBC News, 2009). Scarce resources may be mitigated by state presence, which should have governance structures and mechanisms in place to help the population. Property regulation is one such an example. In areas that are very developed, people will have more coping abilities in place to deal with issues surrounding water access because the state plays a mitigating role. How well a population can cope with water scarcity is dependent on the development within an area, which is dependent on state presence and installed risk reduction mechanisms, infrastructural development, and so on (Martin-Ferrer, Vernaccini, & Poljansek, 2017). A higher state presence should thus at least weaken the effect water scarcity has on communal conflict. Therefore:

H(II). State presence decreases incidences of communal conflict in areas where water is

scarce.

H(IIa). State presence increases the population’s coping capacity which decreases

(19)

Violence is seen as a last resort and the consequences of scarcity are often mitigated by institutions both of formal and informal nature. However, tools to help communities are not infinite and often fail to provide full remedy. A tool of adaptation could be a shift to forms of agriculture that are less water-intense (Döring, 2020), though such measures are often

temporary and do not provide lasting relief. Seeing as many livelihoods are dependent on water to make a living, they will have to find other ways to mitigate their losses. Because of this, these communities and groups will either have to move or obtain access to water sources in another way. In doing so there is an increased risk of violence and intercommunal clashes as people will turn to the same water sources, so:

H(III). Groups whose livelihood is highly dependent on water access have an increased risk

of being part of communal conflict in areas in which water is scarce.

Another factor to consider is the increasing rate of population density globally. Even in towns and villages this may play a role and could lead to increased exploitation of water resources. This has happened drastically in Africa in the past 20 years. Restrictions to water usage have become a problem and pressure on the resources has increased fast, including the higher sewage system demand. This also creates more reason for conflict between farmers and herders considering their need to produce higher outputs (Döring, 2020). This means the denser the population, the more likely water scarcity is going to increase competition and tensions and subsequently lead to communal conflict. Thus:

H(IV). The higher the population density in an area in which water is scarce, the more likely

the risk of communal conflict in this area.

Chapter 3: Methodology

This paper investigates how water scarcity affects the rate of communal conflict in the Horn of Africa. Up until this point, linking communal conflict to water scarcity has been based mainly on precipitation data (Döring, 2020). A broader analysis of water scarcity, however, has indicated the importance of groundwater in determining water availability. To be able to study the link between water scarcity and communal conflict, rather than gathering

(20)

data on the interstate level as previous studies have done, data will be collected in the form of geo-references, focussing specifically on non-state conflict. Having narrowed down to the three countries of Kenya, Somalia, and Ethiopia, in addition to analysing communal violence incidences per country, migration mechanisms will be measured since the borders are left intact in this form of analysis.

This study will build on a causal process tracing (CPT) approach. It is a within-case method concentrating on causal mechanisms linking causes and effects within specific cases (Blatter & Haverland, 2012). Observations made using this method give an insight into the context, process, or mechanism and allow to make a causal inference on the basis of this information. Using CPT means collecting information that can be used to determine the causal conditions, the temporal order in which they cooperate to produce an outcome, whether these conditions are necessary or sufficient, and it also helps identify which mechanisms form the basis of explanations related to the outcome (Blatter & Haverland, 2012). This method starts with the assumptions that an outcome usually results from a number of causal conditions combined, that there are various pathways to such an outcome, and that different contexts and combinations determine the effect of the causal conditions (Blatter & Haverland, 2012). As such, this research will not speak of ‘variables’, but of ‘causal conditions’. Its aim is to find the ‘necessary’ and ‘sufficient’ conditions when determining the outcome, we are trying to investigate (Beach & Pedersen, 2019), namely communal conflict in the Horn of Africa. This also implies that whatever the outcome, it is not generalizable to a wider population of cases (Blatter & Haverland, 2012). However, this research still contributes to the debate by showing which causal pathways and configurations may create a certain outcome. CPT will help increase the internal validity of the causal configuration it tries to establish and helps to identify links between dependent variables and an outcome. Based on the theoretical framework, a possible causal conjunction can be

established between all the dependent variables and the outcome, water scarcity (BBC News, 2009). This should help establish the link between water scarcity and violent conflict in the Horn of Africa and provide a ‘comprehensive storyline’ to present potential causal

mechanisms and develop a potentially relevant mechanism. The main goal of this process is to identify the links and their sequences, to differentiate between them, and to visualize the process leading to the outcome (Blatter & Haverland, 2012). In doing so, the study will provide more detail about the causal processes that occur at different places on different moments, and gather data to support strongly that a causal condition, or a combination of such, actually created the causal pathway and triggered the next step in the process (Blatter &

(21)

Haverland, 2012). It is therefore also considered how different insights from actors or parties may be involved in the process, for example by analysing humanitarian reports, as to not draw inferences purely for temporal reasons (Blatter & Haverland, 2012).

The outcome of communal conflict will thus be based on a model of causation, or a causal pathway, based on several causal conditions. This means that communal conflict is determined by the interaction of the hypothesized causal mechanisms and the conjunction of them. CPT will be used to help establish the complicated chain between conditions leading to violent conflict and set up a causal configuration between water scarcity and communal conflict as such. It will help us understand how communal conflict emerges in relation to water scarcity, and thus help carry out a temporal analysis (Kay & Baker, 2015). Next to previously executed comparative case studies and large-N designs, CPT is a strong complement to help understand the causal mechanisms between variables and outcomes. Using CPT to disaggregate the causal process surrounding water scarcity and communal conflict allows us to analyse in detail all the elements it consists of (Kay & Baker, 2015). The aim of CPT, then, is to hypothesize a causal mechanism and uncover the traces of it within the history of cases (Kay & Baker, 2015). This will be done by using four different types of evidence known to be very important to any process-tracing method: pattern, sequence, trace, and account. Pattern evidence refers to evidence that takes the form of predictions of

statistical patterns. Sequence evidence refers to the analysis of the chronology of events predicted by a hypothesized causal mechanism both spatially and temporally. Trace evidence is used to prove that the parts of a specific hypothesized mechanism exist on the basis of records. Account evidence deals with empirical data content, providing a detailed analysis of the data (Beach & Pedersen, 2012).

The aim of the research is to build theory using inductive reasoning, thus to uncover what the causal mechanism between X, water scarcity, and Y, communal conflict is, based on the different types of empirical evidence (Beach & Pedersen, 2019). This is because there has been evidence of an existing correlation between the two, but there is also uncertainty

surrounding this correlation and as to what causes communal conflict (Beach & Pedersen, 2019). In short, theory-building causal process tracing thus refers to the collection and use of historical, temporal data to build a hypothesized causal conjunction between several causal conditions and an outcome with the aim of developing a theoretical explanation out of this empirical evidence. For this reason, the research question is aimed at finding out what the causal mechanism between water scarcity and communal conflict in the HOA, if any exists at all, hence the use of the word ‘how’. To build such a theoretical explanation, data will be

(22)

collected to build a hypothesized causal conjunction (Kay & Baker, 2015). Data will be traced back to make observations related to communal conflict and water scarcity, and the interaction between those and other causal conditions within this configuration. The data collected will inform the answer to the research question by shedding a light on the processes going on within the causal conjunction and will help analyse and establish the links between the several causal conditions.

Communal Conflict

The Codebook Uppsala Conflict Data Program (UCDP) Non-state Conflict Issues and Actors Dataset formally denotes the UCDP definition of non-state as:

The use of armed force between two organized armed groups, neither of which is the government of a state, which results in at least 25 battle-related deaths in a year. (von Uexkull & Pettersson, 2018, p.2)

Arms here refer to any material that can result in death, from sticks to weapons. By 25 battle-related deaths, the report refers to deaths that are directly battle-related to such use of arms between the involved parties. Organized groups can be either formally organized groups that are non-governmental but have an announced name and use arms against a group organized in a similar manner or informally organized groups that do not have an announced name yet still use violence against a group similarly arranged by using violence. This violence needs to be determined to take place in a clear pattern of connected incidents in which armed force is used by both groups against one another (von Uexkull & Pettersson, 2018).

Following the same definition, a state is either a sovereign in control of a specified territory who has been internationally recognized or an unrecognized government in control of a specified territory, without having had an internationally recognized sovereign previously ruling in that same territory dispute their sovereignty. A government is defined as to be the party in control of the state’s capital (von Uexkull & Pettersson, 2018).

To measure the dependent variable communal conflict, the yearly rate of fatalities as a result of such conflict will be measured, seeing as there have been some ongoing conflicts,

especially in Somalia, that cannot be measured as one conflict whilst still generating so many fatalities. The data concerning the number of deaths from such communal conflicts will be obtained from the Uppsala Conflict Data Program, who present a Geo-referenced Event Dataset (UCDP GED), which has registered nonstate conflicts around the world around 70% of which were registered in Africa alone (von Uexkull & Pettersson, 2018). This will be

(23)

combined with the UCDP Non-State Conflict dataset to achieve higher levels of detail and information on the actors (Döring, 2020). The UCDP GED provides the exact coordinates, latitude and longitude, of the events, describing what has happened with references to news articles and publications.

Previous research has shown that the data shows the largest determinant of communal conflict is territorial conflict, in which agriculture or water is the cause of many issues (Döring, 2020). This UCDP dataset, therefore, represents very broadly the types of

communal violence this research is interested in. From this data, a subset will be created for the HOA, covering the period of 1990 to 2014, limited by data availability. Ethiopia,

Somalia, and Kenya are amongst the African countries with most violent communal conflicts (Döring, 2020). The UCDP Non-state Conflict Issues and Actors Dataset builds upon the Non-state Conflict dataset and provides information about armed nonstate conflict issues and their key actors. This allows for further elaboration into the causes, dynamics, and

consequences of such nonstate violent communal conflicts (von Uexkull & Pettersson, 2018). The dataset includes three organizational levels. The first, formally organized groups, contains rebel groups and other groups that can be categorized in the state-based armed conflict category. The second, informally organized groups, categorizes supporter groups and political party candidates and affiliates. The third level, informally organized groups, are groups commonly identifying on the basis of ethnic, clan, religious, national or tribal

affiliations, or ‘communal conflicts’ (von Uexkull & Pettersson, 2018). Because this research is focused on the outcome of communal conflicts, only incidences falling under

organizational level 3 will be filtered out. Furthermore, as the countries where fighting took place within a given year were coded within the UCDP dataset in line with the country codes per country that were provided by Gleditsch and Ward (2007), the countries were filtered out by their codes: the code 530 within the UCDP dataset refers to Ethiopia, 520 refers to

Somalia, and 501 refers to Kenya. The data for these countries was filtered out and selected to form a separate dataset.

The variables most important to the research have been chosen and are visually

represented in figures 1 to 16 below. Under the variables Side A and Side B Livelihood both communal group parties of the conflict are divided up into the categories agropastoralist, pastoralist, farmer, or other, numbered 1 to 4. The variable Dyadic Livelihood shows the livelihood of both groups combined, in numerical order (von Uexkull & Pettersson, 2018). This variable is important to understand the background of the groups we are studying and might be able to explain their motivation behind use of arms. The dataset has measured a

(24)

total of 134 conflicts divided over the three countries this research focuses on, of which 51 took place in a grid cell within Ethiopia, 37 occurred in Somalia, and 46 were registered in Kenya.

Figure 1 below provides an overview of the livelihoods of both sides, A and B, involved in the conflict. Pastoralism here is defined as nomadic animal husbandry, raising animals such as cattle, goats, sheep, and camels and caring and tending them whilst moving the herd (Meier, Bond & Bond, 2007). Agropastoralism is defined as the combination of farming, which refers to dryland agriculture or the cultivation of soil to grow crops, with the pastoralism as the rearing of animals in order to produce wool, food, and other products (Turner, Ayantunde, Patterson & Patterson, 2012). The category ‘other’ entails any party involved with conflict that does not define as pastoralist, agropastoralist or farmer, or whose livelihood is unknown. A confrontation between two agropastoralist groups then means a confrontation between groups that combine both agricultural farming and the herding of animals. An agropoastoralist-farmer confrontation means a confrontation between such a group and a group that practices farming by means of cultivating soil and growing crops.

Figure 2 illustrates the number of conflicts per conflict issue category per country, which is important to understand the origin of the conflict. The defined conflict issues are territory, authority, and other. Territory contains all territorial issues concerning a piece of land, such as borders, grazing land, wells, or arable land. Authority issues are those conflicts related to the state’s formal administration, such as local administration control, and informal

Figure 1. A visual representation of the livelihoods of both sides A and B involved in a communal conflict based

(25)

control such as ethnic or rebel group leaders. Other issues are issues that see conflict reasons other than territory and authority (von Uexkull & Pettersson, 2018).

Territorial sub-issues are agricultural land and/or water conflict issues which entail conflicts in which water or agricultural lands are the main reason for escalation and territory conflict issues which escalated for other reasons than agricultural land or water. For

agricultural conflict issues, sub-issues may be formal authority issues such as control state apparatus or certain representatives of the government, or informal authority issues in the form group or community leadership. Within the ‘other issues’ category fall the sub-issues of

Figure 2. A bar chart showing the number of occurrences of main territorial conflict issues expressed in number of

conflicts in which these issues played a role per country based on the UCDP Non-State Conflict Dataset by value 1990-2014. Reproduced from von Uexkull and Petterson (2018).

Figure 3. A visual representation of occurrence of the sub-issues expressed in percentage of total

conflicts per country based on the UCDP Non-State Conflict Dataset by value 1990-2014. Reproduced from von Uexkull and Petterson (2018).

(26)

religion, which is the case when at least one side of the conflict has made explicit religious references in their demands, and other sub-issues that neither concern religion nor livestock. Livestock issues regard issues concerning farm animals or herded animals used to produce labour and commodities such as milk and could entail issues with regards to their feeding or placement (von Uexkull & Pettersson, 2018). Figure 3 above shows the division of conflicts divided into these sub-issues in percentage per sub-issue per country. The chart shows the percentage of conflicts that were placed in the category of a particular sub-issue. For Ethiopia, for example, this means that 36% of the conflicts that took place were related to territorial issues concerning territory that is not related to water or agricultural land but instead entails clashes between two groups claiming the same land. Agricultural land and/or water issues are based on the competition surrounding these factors, and also make up a big percentage of sub-issues. For our CPT approach, this means being able to identify the origin of the issues, and helps understand the motivations of the sides involved in the conflict. With regards to the causal conjunction, this will help see which factors are the most prominent determinants of issue escalations, as it forms the starting point when trying to understand what leads to communal conflict. This is an important first step in uncovering the role of water scarcity in relation to such conflict and the conditions and mechanisms that contribute to the outcome.

Water Scarcity

Seeing as water scarcity depends on the availability of all water sources, measures of precipitation, drought, groundwater availability, and the presence of large surface waters will be incorporated in this research. Because the unit of analysis is subnational as we are looking at non-state, communal conflict, this research will use georeferenced data so that it is suitable to study communal conflict on such an intrastate level. An observation will, therefore, be made of communal violence incidences per PRIO-GRID cell, based on the UCDP data by von Uexkull and Petterson (2018), but also on the PRIO-GRID publications for drought and precipitation. Such grid cells provide boundaries insensitive to time. PRIO-GRID cells are based on a GIS shapefile with cells being approximately 55 by 55 km (Tollefsen, Strand & Buhaug, 2012). This means they can even cover temporary farmer mitigation and

(27)

For precipitation, a yearly average map (Figure 4) was generated using the GPCP v2.2 Combined Precipitation Data Set. This dataset gives the total amount of precipitation per grid cell per year. It shows the average precipitation for a year in Africa as measured in

mm/month (Huffman, Bolvin & Adler, 2012). Specifically, it gives the yearly total amount of precipitation in millimetres for each cell, based on the Global Precipitation Climatology Centre’s monthly meteorological statistics. Such monthly data accounts for the intra-annual variability. This means that in the HOA, precipitation figures are low and have not reached

very high within the given time period. In fact, on average, precipitation did not reach higher than 600 mm/year in the grid cells with the highest precipitation levels but indicates most grid cells have a precipitation amount between 0.68 and 300 mm per year (Adler et al., 2003). This figure will help determine whether precipitation may have plaid a role as a causal factor within the conflict conjunction.

Figure 4. Precipitation map based on the GPCP v2.2 Combined Precipitation Data Set by value 1990-2014

(Adler et al., 2003). Precipitation in mm/year 0.68 3232.75 2155.6 1078.45 Ethiopia Somalia Kenya

(28)

Drought will be analysed using a generated map of ‘Proportion of the year experienced drought’ based on SPEIbase data (Figure 5). The Standardized Precipitation

Evapotranspiration Index (SPEI) takes into account both precipitation and potential

evapotranspiration to measure drought (Beguería, Vicente-Serrano, Reig, & Latorre, 2014). The database measures the severity of drought for each cell’s rainy season. The map shows the average monthly precipitation from 1990 to 2014, ranging from low to high drought index. Within the HOA we observe relatively high levels of drought especially in Ethiopia, and Somalia, with some grids showing more moderate to low drought levels. Kenya appears to see less drought than the other two countries, however, in the north does show higher drought levels.

Temperature will be controlled for using GHCN CAMS dataset, a global monthly land surface temperature analysis based on the Global Historical Climatology Network version 2 and the Climate Anomaly Monitoring System to achieve the most accurate data collection (Fan & van den Dool, 2008). Excluding temperature would mean a bias towards a model based on precipitation (Döring, 2020). Figure 6 below illustrates the average temperature in Africa, zoomed in on the Horn of Africa, for the specified time period. It shows that the

SPEI Drought Index

Figure 5. Drought based on the Global SPEI Database, SPEIbase, showing the severity of the drought per each cell by value

(29)

average temperatures in Ethiopia, Kenya, and Somalia have mainly reached between 23.93 and 39.91 degrees Celsius (Fan & van den Dool, 2008). Towards the west of Ethiopia, cooler temperatures can be observed. In combination with the data collected on precipitation and drought, temperature will give an insight into the interaction between a dry climate and the other conditions contributing to water scarcity and help assess to what extent these factors actually do contribute to it, and how, to incorporate them into the hypothesized causal conjunction as an explanation behind water scarcity and its relation to communal conflict.

For groundwater, a georeferenced map (Figure 7) by Döring (2020) will be used that shows the average depth to groundwater in meters. This map also shows surface waters. In accordance with World Bank estimates, to supply water in small-towns, boreholes can be used in small towns in Africa that reach up to 250 metres in depth. Villages typically have wells that reach until 50 metres in depth. This means that even when groundwater depth is higher, there is still a way to access water by means of different wells (Döring, 2020; World Bank, 2018). In Somalia, a lot of wells have been constructed manually even while it has known civil war for years on end and its water resources have been managed poorly. Water holes that run deep enough are found throughout Somalia regardless and form an important buffer during droughts, both in rural and urban regions. Such boreholes tend to reach a depth

Temperature in degrees 39.9 23.9 7.95

Figure 6. Average temperature in the Horn of Africa, Africa, for the time period 1990-2014 in degrees

(30)

between 15 and 200 metres, but some even reach as far as 450 metres (FAO-SWALIM, 2012. In Döring, 2020). Water availability means that people will extract it, but its scarcity and depth can create a large burden for communities. The access to groundwater was captured using several models using to represent the average depth in meters to groundwater. The variables used represent the averages for given grid-cell years, and thus includes communities both where water is deep and where it can be found closer to the surface. The threshold for 75 metres, does not mean that the water is only extracted at this level, but that there must be a considerable amount of groundwater available within this grid cell for wells to be hand-drilled. The data were simulated and processed with PRIO-GRID. The higher the groundwater values, the more limited groundwater access per given grid cell-year. The depths show whether water can be manually extracted or whether this requires drilling. The cost of such drilling is high and, therefore, makes deeper boreholes more inaccessible, thus

Figure 7. Communal conflict incidences and Depth to Groundwater in metres in the Horn of Africa by value

(31)

leaving water scarcer. It is thus assumed that groundwater accessibility is enhanced in areas where depth is lower (Döring, 2020).

Surface waters are also shown on the map in Figure 7 as another non-time-variant measure. Determining the presence of large water surfaces in the form of lakes is done to see if the region intersects with any large rivers or other water bodies that water could potentially be extracted from. Minor water bodies will not be assessed due to the difficulty in capturing their data and the spatial coverage of the research. On the basis of this we can conclude that Kenya has access to Lake Victoria in the southwest, and access to Lake Turkana in the northwest. Furthermore, Ethiopia has access to the Tana Lake in the north and has several smaller lakes running through its centre (Döring, 2020): Zway Lake, Koka Lake, Langano Lake, Abijata Lake, Lake Shala, and the Awasa Lake. This data will help further in

establishing the interactions between causal conditions contributing to water scarcity and its relation to conflict.

Thus, by analysing the data on precipitation, drought, temperature and groundwater and surface water access in combination with conflict occurrence and the livelihoods of conflict groups, with the underlying issue categories of territory, authority and other issues and their sub-issues, a causal inference will be made establishing a hypothetical causal configuration between these conflict origins and water scarcity. This will then be used to further develop a causal mechanism between water scarcity and communal conflict to see how water scarcity actually may lead to conflict. To do so, however, requires the assessment of other causal conditions that influence and mitigate the effect water scarcity has on the population and controls for other conditions that interact with communal conflict.

State presence, IDPs, and population density

The ability of a population to cope with water scarcity is dependent on the development and state presence within a given area, as defined by the Index for Risk Management

(INFORM) (Martin-Ferrer, Vernaccini, & Poljansek, 2017). A population’s vulnerability to threats is determined mainly by their development, by the level of inequality, and thus also by the level of excluded groups (Martin-Ferrer, Vernaccini, & Poljansek, 2017). In line with this, coping capacity, or a lack thereof, is dependent on state involvement in the form of risk reduction mechanisms and other forms of governance, communication, physical

infrastructure, and health system access (Martin-Ferrer, Vernaccini, & Poljansek, 2017). In line with the theoretical framework, communal violence and its relation to water scarcity are

(32)

explained partially by the state’s capacities, which determines how well the population can cope in case of a disaster, and local displacement patterns, and population density. State presence also includes the state’s ability to provide goods to the region, conflict resolution bodies it may have set up, and scarcity mitigation (Döring, 2020). An accurate measure of state presence thus is the level of development within a region. PRIO-GRID allows for an analysis of state presence building upon the data from the Defence Meteorological Program Operational Linescan System (DMSP-OLS) Nighttime Lights Time Series Version 4,

measuring average light emissions overnight per grid cell by detecting visible and near-infrared emission sources at night. This data shows stable lights and provides a digital brightness value for the lights that are detected, giving an insight into how commerce, resource consumption, and the population in general are distributed. The assumption here is that the night-time lights are associated with urban processes and as such their data has been used to map economic activity and water use amongst others (Elvidge, Hsu, Baugh & Ghosh, 2014).

One of the ways in which public goods provision is visible is through an increase in projects concerning infrastructure. Research has proven that the night-time lights are indeed a

Light index 0 0 0 – 0 0 – 0.02 0.02 – 0.99

Figure 8. Average night-time light emission by value 1990-2014. Reproduced from PRIO-GRID (Tollefsen, Strand

(33)

highly reliable measure in this regard. Even in rural areas, such measures can be translated into state intervention and presence. Such night-time lights emissions data can also be used when accounting for state capacity and wealth on the local level, and thus is a reliable sub-national measure (Döring, 2020). Another measure of physical infrastructure is the travel time from a grid cell to the nearest major city, which points to development in terms of urbanization (Martin-Ferrer, Vernaccini, & Poljansek, 2017). To analyse this, data from PRIO-GRID concerning the average travel time in minutes to the nearest major city from a cell area will be visualized and interpreted (Figure 9). Most cells score in the category of 164.73 minutes to 934.06 minutes, however, a closer look at the data shows most cells indicate a value of 500 minutes or higher (Tollefsen, Strand & Buhaug, 2012).

Another factor that will be controlled for is the fact that the level of development within a country, region, or area decreases violence rates and increases the ability for

mitigation of problems, including those concerning water scarcity. Figure 10 shows the GDP per capita for each of the countries, using data from the World Bank, which defines the GDP per capita as a development indicator (World Bank 2020). Analysing the GDP per capita for

Travel time in minutes 29.04 – 164.73 164.73 – 934.06 934.06 – 5296.44 5296.44 – 30032.62

Figure 9. An overview of average travel times to the nearest major city from a cell area, reproduced from PRIO-GRID

The Last Drop: how Water Scarcity Leads to Communal Conflict in the Horn of Africa

The Last Drop: how Water Scarcity Leads to

Communal Conflict in the Horn of Africa

The Last Drop: how Water Scarcity Leads to

Communal Conflict in the Horn of Africa

Table of Contents

Chapter 1: Introduction

Chapter 2: Theoretical framework

Chapter 3: Methodology