The Role of Ethnicity in Civil Conflicts

2.1 Ethnicity and Ethnic Identity: What Are They?

What is ethnicity? How do we know that there is a distinct ethnic group? Though observers normally have few problems identifying ethnic groups in everyday life, a scientific definition proves more difficult because two basic concepts are opposed. According to an essentialist or primordialist notion of ethnicity, ethnic group identities are stable over time and result from differences in shared identity markers such as clan, community, faith (religion), language, regional provenance, race, sect or tribe, and complexion or other physical appearance (Fearon/ Laitin 2000: 849). However, there is no fixed set or number of such “objective” differences. Sometimes many of these different identity markers are present, but often they are not. [1]

Contrary to essentialists, constructivists argue that ethnic identity can change over time and, at the end of the day, it is the result of self‐ascription and/or ascription by others (Horowitz 1985; Fearon/Laitin 2000: 851‐853). Ethnicity is socially constructed. Yet, it does not seem plausible that ethnic identity changes on a daily or arbitrary basis and is completely independent from “objective” identity markers, though their number and importance may differ from case to case. [2]

Hence, it seems reasonable to define ethnicity in a combined concept, but one that includes more constructivist ideas than primordialist ones (see also Horowitz 1985; Kasfir 1976 [3]): ethnicity and ethnic identity derive from the notion of a common ancestry and are associated with a variable set of objective identity markers such as language, religion, and physical appearance. However, though relatively stable over time, ethnic identity finally results from self‐ and outside ascription and may be principally subject to change.

2.2 Ethnic Conflicts: What Do We Know About Them?

What do we know about ethnic conflict? To begin with, it seems useful to define it. As it is broadly understood, ethnic conflict denotes any conflict in which at least two ethnic groups are opposed—at least as the main bases of the warring factions—with regard to an incompatibility such as access to power and resources or more symbolic incompatibilities such as discourses on history. Such a conflict can be violent or peaceful. [4]

While many economic, social, and political phenomena may impact ethnic violence, a minimum of ethnic diversity is a precondition for the onset of ethnic violence. But why might ethnic diversity result in interethnic violence? Theoretically, there is a number of ways (see Fearon 2003; Hoeffler, forthcoming; Bussmann/Hasenclever/Schneider 2009: 15‐18). Ethnic diversity may be particularly prone to (violent) conflict because of socio‐psychological inter‐ group dynamics. A positive in‐group identity is necessarily associated with a more negative perception of the out‐group. Thus, negative action against the out‐group becomes more likely. Ethnically defined in‐ and out‐groups might be particularly prone to such escalation because often—not always, as argued above—the out‐group has many objective differences and its “otherness” can hardly be denied. In a context of economic crisis, individuals tend to seek refuge in ethnic identities (see Bussmann/Hasenclever/Schneider 2009: 15‐18). This will be politically relevant all the more when ethnic groups suffer from (perceived) relative deprivation in economic, political, or other terms (Gurr 1970; 2000). In any case, ethnic identities are resources that leaders can draw on for political mobilization.

The quantitatively oriented empirical research on ethnic conflict has focused on the question of whether and what particular constellations of ethnic diversity are especially conflict‐prone. There are three basic constellations of ethnic diversity: Ethnic fractionalization measures the number of ethnic groups and their relative size. A value of 0 denotes that all individuals belong to the same group, while a value of 1 means that all individuals belong to a different group (Fearon 2003; Collier/Hoeffler 2004). The idea behind this measure is simple: the more diversity, the more conflict. In contrast, Horowitz (1985) argues that it is not higher fractionalization, but so‐called ethnic polarization that is most conflict‐prone (see also Esteban/ Schneider 2008). Polarization does not refer to the quality of the relations—as it may suggest—but to a demographic constellation in which few but bigger groups are opposed, ideally two groups with a 50 percent share of the population each. An index on that (Montalvo/ Reynal‐Querol 2005) measures how close a demographic constellation is to two groups, each with a share of 50 percent. Since in such a constellation the differences between the groups are more salient and hence easier to mobilize, polarization might be more conflict‐prone than fractionalization, in which the large number of groups creates a collective action problem. Finally, a third concept or measure is ethnic dominance, in which one ethnic group has a relative or absolute majority within the population of the country, potentially marginalizing smaller groups. Measures differ as to whether the threshold of dominance stands at 40, 50 or 60 percent of the population share of the group in question.

There are many quantitative studies that have tried to test the different measures and their impact on civil war onset (e.g. Collier/Hoeffler 2004; Fearon 2003; 2005; for an overview, see Hegre and Sambanis 2006; Dixon 2009; Hoeffler forthcoming). To make a long story short: There are no robust results. None of the three measures has proven robust in all the different studies. There is just one consensus: ethnic diversity as such is not necessarily associated with civil war or other forms of organized violence.

Certainly, these measures of ethnic diversity have severe disadvantages (Posner 2004; Cederman et al. 2009). First, their reliability might certainly be questioned. Given the lack of comprehensive worldwide data on the ethnic composition of countries, measures rely most of the time on a Soviet atlas from the 1960s, freezing a picture that has obviously evolved since then. Second, critics have noted that behind identical index values for ethnic fractionalization and polarization, very different constellations can be hiding. [5] Third, it may be argued that it is not so much the quantity but the quality of relations that makes the differences. Different measures such as the “politically relevant ethnic groups” proposed by Posner (2004) or the Minorities at Risk Project may thus better capture the salience and quality of ethnic cleavages or deprivation of ethnic groups. For instance, in a setting of a dominant ethnic group, whether the minority rules or is suppressed (or neither/nor) will differ.

Recent research indeed provides support for the view that the quality of relations in the sense of Gurr ’s relative deprivation thesis matters. A data set compiled by Cederman, Min and Wimmer (2009) on the basis of judgments by country experts has extensively coded the nature of power relations between ethnic groups. Their analysis finds that the political exclusion of ethnic groups is a significant determinant for civil war and ethnic strife (Cederman et al. 2010). This finding seems highly plausible. However, it raises the question of why the exclusion of ethnic groups—or the mobilization and instrumentalization of ethnic identity—happens in some cases while it does not in other cases. Apparently, it is not directly related to specific demographic constellations of ethnicity. Generally, we should be aware that if any measure of ethnicity proves significant, other factors such as economic inequalities and political power struggles among elites also have to be taken into account.

[1] In the case of Hutu and Tutsi in the Great Lakes region, for instance, there are very few objective differences.

Language, religion and regional settlements are identical. However, Tutsi were historically dominant in the pre‐colonial Tutsi Kingdoms, (and the colonial administration maintained and even amplified this Tutsi “supremacy”), and it is the belief that Tutsi descend from Hamitic pastoralists while Hutu were originally sedentary Bantu farmers. The resulting differences in physical appearance—tall and thin Tutsi vs. shorter Hutu— are rarely visible at first glance. The genocidaires in Rwanda in 1994 often had to check the ethnic affiliation on a person’s identity card—where the ethnic identity was noted—before slaying that person.

[2] An additional problem arises from different levels of analysis. Do we look at larger groups or subgroups?

Sometimes differences within subgroups might be politically more relevant than those between larger groups.

[3] For Kasfir, “ethnicity contains objective characteristics associated with common ancestry, such as language, territory, cultural practices and the like. These are perceived by both insiders and outsiders as important indicators of identity, so that they can become the bases for mobilizing social solidarity and which in certain situations result in political activity” (Kasfir 1976: 77).

[4] If not indicated otherwise, ethnic conflict will denote violent ethnic conflict.

[5] For example, three groups of equal size will have the same value, with one group with approximately a 50% population share and a number of smaller groups with less than a 10% population share each.