Final Report Summary - ECCO (Ethnic/Cultural Conflicts and Patterns of Violence)
The ECCO project was implemented at the Peace Research Institute Oslo from August 2010 through September 2012, interrupted for an eight-month research stay at Yale University. The main task of the project was to analyse internal dynamics of ethnic violence, looking both at spatial and temporal patterns of violence. In line with the project schedule, the fellow was able to complete almost all of the project goals for the first year of the project ('national perspective'), and was able to make progress also on the second half ('transnational perspective'). The project conducted research along three main perspectives: Why does ethnic violence occur? Where does it occur? and What are its impacts? In addition, the project aimed to make contributions improving the empirical study of ethnic violence more generally. In the following, we briefly summarise the approach and the outcomes of the project for the different perspectives, starting with the national perspective.
Why does ethnic violence occur? Ethnic violence has been attributed to a number of reasons. Some scholars argue that violence is the result of opportunities that arise for it to be employed. These opportunities could exist, for example, in weak states that are unable to keep ethnic challengers in check. While this perspective certainly has some value, it moves the focus away for ethnic grievances as original triggers of conflict. Political discrimination has been shown to be an important cause of violence. In addition, the project conducted research on economic disadvantages and how they relate to political grievances as motive for violence. In order to test this mechanism empirically, the project developed a novel spatial measurement procedure to estimate the level of wealth of ethnic groups. Using these estimates, it was possible to assess whether a particular group is economically disadvantaged in comparison with the national average. These 'horizontal inequalities' were then tested in a regression analysis on a global sample of groups. The result is that both political and economic disadvantages foster conflict, and operate in parallel. In future work, the fellow will extend this work to focus on intra-group inequalities and how they affect conflict risk.
Where does ethnic violence occur? While the onset of ethnic conflict has received a considerable amount of attention in the literature, there is much less research on its spatial patterns. The project conducted an extensive study of the spatial distribution of violence in the Bosnian civil war in 1992-1995. There are two explanations for violence: First, some scholars argue that violence primarily served the purpose of carving out ethnically homogenous territories, in a sense following a top-down mechanism where violence follows a higher order, and is otherwise unrelated to the local conditions. Second, quite the opposite, other scholars attribute violence to local inter-ethnic enmities and feuds. The project developed two indicators that measure both predictors. These indicators were then used in a statistical analysis, which revealed that most of the violence in Bosnia can be explained by the first mechanism, which indicates that the territorial logic of ethnic cleansing was dominant. In a second study, the project analysed the dynamics of violence and segregation at a much finer scale, by looking at the case of a single city. Using Baghdad as a case, the project created a computational model, where inter-ethnic violence generates fear in the population, which then leads to migration. The model was validated using empirical data at the neighborhood level, and constitutes one of the first attempts in political science to use a geo-referenced computational model built on real-world empirical data.
What are the impacts? Last, the project conducted a micro-level study of the consequences of violence. Using data collected in a survey in Northern Afghanistan, the project combined these data with individual reports of where and when violence occurred. In particular, the study was interested in how violence nearby impacts social cohesion at the village level. The results reveal that violence has a negative effect on the social glue, it makes villages diverge more in terms of trust and cohesion. Related to this particular question, the project also conducted research on the impact of violence on democratisation efforts, looking in particular on election fraud. A study conducted on the 2009 presidential election in Afghanistan revealed that violence benefits fraud, but only up to a certain degree. Once violence becomes too intense, even the perpetrators of fraud will stop operating, which leads to the interesting finding of low fraud in high-violence regions.
The project also conducted research at the transnational level, by analysing the spread of ethnic conflict from one country to another. This was done by creating a stylised computational model, where social norms of ethnic dissimilationin other words, the common perception that ethnic fault lines are social relevanttravel from one country to another. While this is only one mechanisms of how conflict can spread, it is one that has been frequently mentioned in the literature, but has received little systematic analysis. Using a computational model of two countries each populated with a number of citizens, the models tests a top-down, entrepreneur-driven mechanism of norm diffusion against a bottom-up, society-driven one. The results reveals that the first one is more robust to social complexity and imperfect information, and thus constitutes a more likely candidate explanation for the spread of ethnic dissimilation.
Why does ethnic violence occur? Ethnic violence has been attributed to a number of reasons. Some scholars argue that violence is the result of opportunities that arise for it to be employed. These opportunities could exist, for example, in weak states that are unable to keep ethnic challengers in check. While this perspective certainly has some value, it moves the focus away for ethnic grievances as original triggers of conflict. Political discrimination has been shown to be an important cause of violence. In addition, the project conducted research on economic disadvantages and how they relate to political grievances as motive for violence. In order to test this mechanism empirically, the project developed a novel spatial measurement procedure to estimate the level of wealth of ethnic groups. Using these estimates, it was possible to assess whether a particular group is economically disadvantaged in comparison with the national average. These 'horizontal inequalities' were then tested in a regression analysis on a global sample of groups. The result is that both political and economic disadvantages foster conflict, and operate in parallel. In future work, the fellow will extend this work to focus on intra-group inequalities and how they affect conflict risk.
Where does ethnic violence occur? While the onset of ethnic conflict has received a considerable amount of attention in the literature, there is much less research on its spatial patterns. The project conducted an extensive study of the spatial distribution of violence in the Bosnian civil war in 1992-1995. There are two explanations for violence: First, some scholars argue that violence primarily served the purpose of carving out ethnically homogenous territories, in a sense following a top-down mechanism where violence follows a higher order, and is otherwise unrelated to the local conditions. Second, quite the opposite, other scholars attribute violence to local inter-ethnic enmities and feuds. The project developed two indicators that measure both predictors. These indicators were then used in a statistical analysis, which revealed that most of the violence in Bosnia can be explained by the first mechanism, which indicates that the territorial logic of ethnic cleansing was dominant. In a second study, the project analysed the dynamics of violence and segregation at a much finer scale, by looking at the case of a single city. Using Baghdad as a case, the project created a computational model, where inter-ethnic violence generates fear in the population, which then leads to migration. The model was validated using empirical data at the neighborhood level, and constitutes one of the first attempts in political science to use a geo-referenced computational model built on real-world empirical data.
What are the impacts? Last, the project conducted a micro-level study of the consequences of violence. Using data collected in a survey in Northern Afghanistan, the project combined these data with individual reports of where and when violence occurred. In particular, the study was interested in how violence nearby impacts social cohesion at the village level. The results reveal that violence has a negative effect on the social glue, it makes villages diverge more in terms of trust and cohesion. Related to this particular question, the project also conducted research on the impact of violence on democratisation efforts, looking in particular on election fraud. A study conducted on the 2009 presidential election in Afghanistan revealed that violence benefits fraud, but only up to a certain degree. Once violence becomes too intense, even the perpetrators of fraud will stop operating, which leads to the interesting finding of low fraud in high-violence regions.
The project also conducted research at the transnational level, by analysing the spread of ethnic conflict from one country to another. This was done by creating a stylised computational model, where social norms of ethnic dissimilationin other words, the common perception that ethnic fault lines are social relevanttravel from one country to another. While this is only one mechanisms of how conflict can spread, it is one that has been frequently mentioned in the literature, but has received little systematic analysis. Using a computational model of two countries each populated with a number of citizens, the models tests a top-down, entrepreneur-driven mechanism of norm diffusion against a bottom-up, society-driven one. The results reveals that the first one is more robust to social complexity and imperfect information, and thus constitutes a more likely candidate explanation for the spread of ethnic dissimilation.