Sammendrag
The majority of existing studies of internal armed conflict have explored the causes of conflict at the country level. While factors such as democracy, composition of the ruling elite or memberships of military alliances are best studied at the national level, other factors vary more considerably within the territory of a state - factors such as rugged terrain, socio-economic development and poverty, quality and presence of institutions, environmental conditions or the location of excluded minorities.
Through five independent but related articles, this thesis explores how local factors such as physical and socio-cultural inaccessibility, local poverty and the quality of local institutions affect the sub-national risk of conflict. It also examines whether recurring conflicts have different severity levels than initial conflicts and how the outcome of the initial conflict affects the severity of the subsequent conflict.
The first article, “PRIO-GRID: A unified spatial data structure,” co-authored with Halvard Buhaug and Håvard Strand and published in Journal of Peace Research, describes the development and application of a spatial framework for the disaggregated study of civil war. It emphasizes the importance of disaggregated research designs and the increasing use of georeferenced data. The dataset consists of sub-national data at the grid cell level, on a large selection of political, economic, demographic and environmental variables covering the period 1946-2014.
The second article, “Insurgency and inaccessibility,” co-authored with Halvard Buhaug and published in International Studies Review, examine the role of physical and socio-cultural inaccessibility in shaping the opportunity and motivation for local armed conflict. The article employs sub-national data on inaccessible terrain and politically excluded groups. The findings demonstrate that inaccessibility is a central factor affecting local conflict risk, where remote areas are significantly more conflict prone than more accessible parts of a country.
The third article, “Experienced Poverty and Local Conflict Violence,” single-authored, analyze geo-referenced survey data for 4,008 districts, across 35 African states, and finds that impoverished areas are more likely to experience conflict violence. However, the article finds a strong pacifying effect of the quality of local institutions. Poverty is only related to conflict if local institutions are weak and ineffective. The results also show that poverty is more strongly associated with conflict if group grievances exist locally than if local groups perceive themselves as fairly treated by the government.
The fourth article, “Local institutional quality and conflict violence in Africa,” co-authored with Tore Wig and published in Political Geography, combines georeferenced survey data from the Afrobarometer surveys with georeferenced conflict data. The results show that administrative districts with high-quality local government institutions are less likely to experience violence in an internal conflict than poorly governed districts. This relationship holds when controlling for country-level institutions and other characteristics.
The fifth article, “Every Cloud has a Silver Lining: The Severity of Armed Conflict Recurrences,” single-authored, examine whether subsequent conflicts are more or less deadly than initial conflict episodes, and if so, what explains these differences. By using data on internal armed conflict episodes between 1946 and 2014, the results show that recurring conflicts are almost 50 % less deadly than initial conflicts. Previous studies suggest that conflicts that terminate in settlements are more likely to recur. The results presented herein shows that once peace agreements fail, the result is a less severe conflict recurrence, than if either party prevailed.
This thesis contributes to a geographical understanding of the local causes of armed conflict in two ways. First, it contributes to shifting away from the state-centric understanding of conflict, or a “territorial trap,” through spatial disaggregation from the homogeneous state to the local. Second, this thesis does not take an absolute understanding of space, where units are only separated by Euclidean distance, but emphasizes the relative and relational connectivity between individuals and institutions, as well as between the local and the national.