Editor’s note: This column updates the authors’ earlier column, Michalopoulos and Papaioannou (2012).
"We have been engaged in drawing lines upon maps where no white man's feet have ever trod; we have been giving away mountains and rivers and lakes to each other, only hindered by the small impediment that we never knew exactly where the mountains and rivers and lakes were"
– Lord Salisbury, British Prime Minister, 1885-1892 and 1895-1902.
The predominant explanations on the deep roots of contemporary African development centre on the influence of Europeans during the colonial period (Acemoglu et al. 2005), but also in the centuries before colonisation when close to 20 million slaves were exported from Africa (Nunn 2008).1 Yet, another milestone took place amidst these two major events - that according to the African historiography (Herbst 2000) had malicious long-lasting consequences. In a forthcoming paper (Michalopoulos and Papaioannou 2016), we examine the impact of the ‘Scramble for Africa’ (1860-1905), when Europeans partitioned Africa into spheres of influence, protectorates, and colonies. The borders were designed at a time when Europeans had barely settled in Africa and had limited knowledge of the local conditions. Despite their arbitrariness these boundaries outlived the colonial era. Jeffrey Herbst summarises: "[f]or the first time in Africa's history [at independence], territorial boundaries acquired salience...The boundaries were, in many ways, the most consequential part of the colonial state."2 While African scholars have presented case-study evidence on the negative consequences of the improper border design (see Dowden 2008 for an eloquent narrative), there is little work formally examining the impact of ethnic partitioning focusing on the universe of African ethnicities. Our work is a first step in this direction.
Identifying partitioned ethnicities
To identify partitioned groups in a systematic manner, we project contemporary country borders to George Peter Murdock's Ethnolinguistic Map (1959) that depicts the spatial distribution of African ethnicities at the time of colonisation (Figure 1a). We classify as partitioned those ethnicities with at least 10% of their total area falling to more than one country. There are 229 ethnicities (out of 825) with at least 10% of their historical homeland falling into more than one contemporary state (Figure 1b). While there is noise in Murdock’s map, which does not allow for a likely partial overlapping of groups’ and although we do not have good ethnic-specific population estimates, our procedure identifies most major partitioned ethnicities.
A primer on border artificiality
The African historiography provides ample evidence arguing that, in the majority of cases, Europeans did not consider ethnic features and local geography in the design of colonial borders. A.I. Asiwaju (1985) summarises: "[t]he study of European archives supports the accidental rather than a conspiratorial theory of the marking of African boundaries". At the time Europeans had limited knowledge of local conditions, since, with the exception of some coastal areas, the continent was largely unexplored. Moreover, at the time Europeans were not drawing borders of prospective states or – in many cases – even colonies. In a few instances, nevertheless, Europeans did try taking into account political geography, as, for example, in Swaziland and Burundi. And few borders were delineated in the early 20th century (Somalia), when Europeans conceivably had some knowledge of local conditions.
We first examine whether partitioned and non-split ethnic homelands differ with respect to various geographical, ecological, political and other features, as this may shed light on whether coloniser’s decisions reflected such factors. With the exceptions of the land mass of the historical ethnic homeland and (weakly) the presence of lakes, there are no significant differences between split and non-split homelands along a comprehensive set of covariates that besides geography-ecology also reflect pre-colonial, ethnic features, such as the size of settlements, the type of the subsistence economy, practice of slavery, etc.
Ethnic partitioning and conflict
We then employ the Scramble for Africa as a ‘quasi-natural’ experiment to assess the impact of ethnic partitioning on civil conflict, as this has been theorised to be the main channel of influence. We exploit on a newly assembled dataset (Armed Conflict Location and Event Data Project) that provides georeferenced information for 1997-2013 on battles between government forces, rebels and militias, one-sided civilian violence, riots, and foreign interventions.
Figure 2a gives the spatial distribution of conflict over this period. The map plots 64,650 high-precision georeferenced conflicts. We conduct the analysis at the historic-ethnic-homeland level, so as to account for national politics, colonial and post-independence institutions, and other country features – the regional analysis allows controlling at a fine level for geography, natural resources, and ecology. Figure 2b portrays the spatial distribution of all conflict incidents at the country-ethnic homeland level.
Our econometric analysis shows that civil conflict is significantly higher in the homelands of partitioned, as compared to non-split, ethnicities. The strong link between ethnic partitioning and conflict is also present, when we restrict the estimation to ethnic homelands close to the national borders, so as to account for any border effects (which nonetheless could be driven by partitioning). Our estimates suggest that conflict intensity is approximately 40% higher and conflict lasts on average 55% longer in the homelands of partitioned ethnicities. And the likelihood of conflict is approximately 8% higher in the homelands of split ethnicities. We also document spillovers to areas adjacent to split ethnicities. Conflict intensity is approximately 30% higher and the likelihood of conflict increases by 7% in the homelands of groups that are surrounded by 50% of split groups (as compared to groups whose neighbours are not partitioned).
Ethnic partitioning and type of conflict
We then exploit the richness of the data to shed light on the mechanisms. First, we examine the thesis put forward by African historians that split groups are often used by neighbouring countries to stage proxy wars and destabilise the government on the other side of the border. Our analysis supports this conjecture. Military interventions from adjacent countries are more common in the homelands of partitioned groups, rather than in nearby border areas where non-split groups reside. Second, we examine the impact of ethnic partitioning on the different forms of political violence. Partitioning matters crucially for two-sided conflict between government troops and rebel groups "whose goal is to counter an established national governing regime by violent acts" and to a lesser extent with one-sided violence against civilians. These patterns are corroborated with a different georeferenced conflict database (Uppsala Conflict Data Program Georeferenced Event Dataset) that records only deadly events associated with major civil wars. In contrast, there is no link between ethnic partitioning and riots/protests, which are predominantly a capital-city phenomenon – and there is no association between partitioning and conflict between non-state actors. These results are in accord with African historiography pointing out that partitioned groups face discrimination from the national government and often engage in rebellions to counter repression.
Ethnic partitioning and political violence
In an attempt to dig deeper on the partitioning-repression-civil war nexus we use the Ethnic Power Relation dataset (Wimmer et al. 2009), which offers an assessment of formal and informal degrees of political power in national politics since independence. The analysis reveals three key findings:
- First, partitioned ethnicities are significantly more likely (11%-14% higher likelihood) to engage in civil wars that have an explicit ethnic dimension.
Seventy-two out of the 234 split groups in our sample (31%) have participated in an ethnic-based civil war, while 69 out of the 359 non-split groups (19%) have participated in a war with a specific ethnic dimension (Figure 3).
- Second, the likelihood that split ethnicities are subject to discrimination from the national government is approximately 7 percentage points higher compared to non-split groups.
Fifty-eight of the 234 split groups have suffered from discrimination (25%), while the corresponding likelihood for non-split groups is 15% (Figure 3).
- Third, we examine the impact of partitioning jointly on one-sided violence (repression) and two-sided violence (civil war).
We find that in the weakly institutionalised African countries, ethnic partitioning leads much more often to major two-sided conflict, rather than on political discrimination; thanks to support from nearby countries and the low opportunity cost of war, repression often escalates into full-fledged conflict.3
Finally, we examine the role of ethnic partitioning on public goods using information from 85,000 households across 20 African countries (Demographic and Health Surveys). Individuals who self-identify with partitioned ethnicities have fewer household assets, poorer access to utilities, and worse educational outcomes, as compared to respondents from non-split ethnicities in the same country (and even in the same town/village). This pattern is not due to a generalised decline in standards of living of all households residing in split homelands – rather, it is driven by the poorer economic performance of members of split ethnicities irrespective of their actual residence.
Our work shows that, by splitting ethnicities across countries, the colonial border design has spurred political violence. Ethnic partitioning is systematically linked to civil conflict, discrimination by the national government, and instability. More research is needed to examine the precise mechanisms at play and elucidate other aspects of state artificiality, related to country size, isolation, and fractionalisation.
And since border artificiality and ethnic partitioning are not phenomena exclusive to Africa, subsequent works perhaps in the Middle East could shed light on how the colonial borders in this part of the world may have a contributed to violence and civil strife.
Acemoglu, D, S Johnson, and J A Robinson (2015), “Institutions as a Fundamental Cause of Long-Run Growth,” in The Handbook of Economic Growth, P Aghion, and S N Durlauf (eds.), Elsevier North-Holland, Amsterdam, Netherlands.
Alesina, A, W Easterly, and J Matuszeski (2011), “Artificial States”, Journal of the European Economic Association 9(2): 246.277.
Asiwaju, A I (1985), Partitioned Africans. The Conceptual Framework, St. Martin Press, New York.
Besley, T and T Persson (2011), “The Logic of Political Violence”, Quarterly Journal of Economics, 126(3): 1411.1445.
Blattman, C and E Miguel (2010), “Civil War”, Journal of Economic Literature, 48(1): 3.57.
Dowden, R (2008), Africa: Altered States, Ordinary Miracles, Portobello Books Ltd, London, UK.
Herbst, J (2000), States and Power in Africa, Princeton University Press, Princeton, NJ.
Michalopoulos, S and E Papaioannou (2014), “National Institutions and Subnational Development in Africa”, Quarterly Journal of Economics 129(1): 151-213.
Michalopoulos, S and E Papaioannou (2016), ”The Long-Run Effects of the Scramble for Africa”, American Economic Review, forthcoming
Nunn, N (2008), “The Long Term Effects of Africa’s Slave Trades”, Quarterly Journal of Economics, 123(1): 139-176.
Wimmer, A, L –E Cederman, and B Min (2009), “Ethnic Politics and Armed Con.ict. A Configurational Analysis of a New Global Dataset”, American Sociological Review, 74(2): 316-337.
1 On the role of Africa’s slave trade on contemporary development, see Nathan Nunn (2008).
2 As a result in many African countries today a significant fraction of the population belongs to ethnicities that are partitioned among different states. Alesina et al. (2011) estimate that partitioned groups constitute on average 40% of the total population.
3 This approach follows Besley and Persson (2011), who proposed studying jointly repression and civil conflict.