National versus Ethnic Identification in Africa: Modernization, Colonial Legacy, and the Origins of Territorial Nationalism Amanda Lea Robinson∗ October 31, 2013 Abstract Communal conflicts, civil wars, and state collapse have led many to portray the no- tion of African nation-states as an oxymoron. Scholars of African politics have argued that strong and persistent ethnic attachments are the result of economic and political modernization, the very forces credited with reducing parochial ties in Europe. Others have argued that national consolidation in Africa is particularly unlikely due to high de- grees of ethnic diversity, colonial rule that exacerbated that diversity, and the partition of cultural groups. Despite the ubiquity of these arguments, there has been very little com- parative, empirical research on territorial nationalism in Africa. Using individual level data from sixteen countries, combined with a novel compilation of ethnic group and state characteristics, I am able to evaluate the observable implications of these long-respected theoretical traditions within a multilevel framework. I find that attachment to the nation, relative to one's ethnic group, increases with education, urbanization, and formal employ- ment at the individual level, and with economic development at the state level { patterns more consistent with classic modernization theory than second generation modernization theory. Thus, if modernization in Africa does indeed intensify ethnic attachment, the impact is overwhelmed by the concurrent increase in pan-ethnic territorial nationalism. Similarly, the results show that ethnic diversity and the partition of ethnic groups by “artificial” state borders increase, rather than decrease, the degree to which individuals identify nationally. Taken together, these results reject pessimistic expectations of African exceptionalism and instead suggest that the emergence of widespread national identifica- tion within territorially-defined African states is not only possible, but increasingly likely with greater economic development. ∗Assistant Professor, Department of Political Science, The Ohio State University. Email: [email protected]. The author is grateful to four anonymous reviewers, James Fearon, David Laitin, members of the Comparative Politics Workshop at Stanford University, and members of the Working Group in African Political Economy (WGAPE) for their feedback on previous versions of this paper. National vs. Ethnic Identification in Africa 1 Weak nationalism is commonly blamed for a host of problems in sub-Saharan Africa, including protracted civil wars, chronic political instability, and economic underdevelopment. As Collier (2009) puts it, \a society can function perfectly well if its citizens hold multiple identities, but problems arise when those subnational identities arouse loyalties that override loyalty to the nation as a whole" (p. 50). The general assumption, then, is that attachment to the state- defined national identity {or \territorial nationalism" (Young, 2004) { is not just weak, but is weak relative to sub-national ethnic attachment. Despite the ubiquity of this assumption, there has been very little comparative, empirical research on territorial nationalism in Africa. This paper makes a first step in that direction by utilizing public opinion data to describe patterns of national relative to ethnic group identification within and across sixteen African countries. By establishing the correlates of greater national relative to ethnic identification, the paper offers insight into the origins of territorial nationalism in Africa. While the empirical study of territorial nationalism in Africa has been rare, theoretically de- rived expectations are abundant. When African states gained independence, there was great optimism that from economic and political modernization, national unity would surely follow. This optimism was founded on classic modernization theories, which argued that processes of \modernization" { including urbanization, universal education, access to mass media, and industrialization { give rise to national identification at the expense of ethnic and other com- munal forms of sub-national group identification (Deutsch, 1953; Gellner, 1964; Weber, 1979; Gellner, 1983; Anderson, 1991). When ethnic group identification did not immediately wane and, in some cases, seemed to grow stronger, a second generation of modernization theory, largely focused on Africa, was proposed. These scholars theorized that the intense compe- tition over resources that resulted from modernization was more likely to engender ethnic identification than to destroy it, resulting in greater fragmentation than national unity (Mel- son and Wolpe, 1970; Bates, 1983). While these two theoretical traditions are not necessarily incompatible in terms of modernization's impact on absolute levels of group identification { it may be that political and economic modernization lead to an increase in both national and ethnic group identification { they do generate competing expectations about the effect that modernization should have on the relative importance of national and ethnic identities in contemporary African societies. A third theoretical tradition rejects the expectation that modernization in Africa will engender national identification on the grounds that African states emerged in a fundamentally different way than the European states on which classic modernization theory was based (Kedourie, 1970; Davidson, 1992). In particular, the territorial borders of African states were carved out by colonial powers without regard for existing patterns of group identification (Herbst, 1989). Partly as a result of the colonial partition, African states are among the most ethnically diverse in the world, with many ethnic groups split into multiple states, producing a context in which national unity could be hard to foster (Young, 1976). Further, the experience of colonial rule by a particular foreign power, as well as the form of anti-colonial struggle experienced, have National vs. Ethnic Identification in Africa 2 also been suggested to account for patterns of group identification (Mazrui, 1983). In short, a diverse literature asserts a form of \African exceptionalism," expecting that the colonial legacy of African states is paramount to understanding variation in contemporary territorial nationalism. To evaluate the impact of modernization and colonial legacies on group identification, this study takes advantage of individual-level survey data on national versus ethnic identification from a representative sample of citizens across sixteen African countries. The use of a rel- ative measure of national and ethnic group identification, rather than a focus on absolute levels, increases the interpersonal comparability of responses, incorporates the constructivist acknowledgement of multiple identities, and is theoretically justified given the competing expectations of the impact of modernization on group identification. Individual level data on relative group identification are combined with a novel compilation of ethnic group and state level data in order to estimate the impact of modernization and colonial legacy on national relative to ethnic identification within a multilevel model. The results lend support to classic modernization theories by showing that living in urban areas, having more education, and being formally employed in the modern sector are all positively correlated with identifying with the nation above one's ethnic group. Further, greater economic development at the state level is also associated with greater national identification, once Tanzania is excluded as an outlier. Thus, these findings are more consistent with classic modernization theories than the expectations derived from second generalization modernization theory. In terms of colonial legacy, the results indicate that purported obstacles to national unity in Africa { highly diverse states and partitioned ethnic groups { are actually associated with higher levels of national versus ethnic identification. I thus conclude that the colonial origins of African states, and the resulting diversity and partition of their societies, has not made them immune to the unifying effects of modernization. This is very important in light of what we know about the impact of increased national identification on rates of interethnic cooperation (Miguel, 2004; Charnysh, Lucas, and Singh, 2013), the promotion of intergroup trust (Robinson, 2013), support for minority-favoring policies (Transue, 2007), levels of economic redistribution (Shayo, 2009), and the likelihood of ethnic conflict (Sambanis and Shayo, 2013). Because strong national identification may have these and other positive implications for African states, it is important to understand the factors that are associated with greater national identification relative to sub-national ethnic identification. These findings bring new data to bear on long-standing debates in the literature on territorial nationalism and ethnic politics in Africa. This is a crucial contribution to the study of terri- torial nationalism, as past studies have often lamented the lack of empirical data on national identification (e.g., Hobsbawm, 1990; Herbst, 2000; Young, 2004). Because the data are at- titudinal, cross-sectional, and come from a limited, non-random sample of African countries, National vs. Ethnic Identification in Africa 3 the results are vulnerable to concerns about social desirability bias, endogeneity, and repre- sentativeness. These