The conventional wisdom on the deep determinants of African development places colonisation at the centre of any explanation. An influential body of research suggests that the extractive colonial institutions that Europeans established in the late 19th century, crucially contributed to African underdevelopment. This channel invokes the lack of checks and balances on the executive and the poorly performing legal systems that endured after African independence (Acemoglu et al. 2005). But European interference in the continent took several other forms.
The slave trades during the 15th-19th century, for example, also hindered its subsequent economic performance (Nunn 2008). Moreover, the artificial border design seems to have played a negative role by increasing the degree of ethnic fragmentation and spurring civil conflict (Englebert et al. 2001, Alesina et al. 2010). Yet some African scholars downplay the importance of colonial and even contemporary institutional structures and stress the role of deeply rooted historically determined ethnic characteristics. Anecdotal evidence, case studies, and a few academic works show that ethnic-specific institutional and cultural characteristics matter, even today (Herbst 2001, Dowden 2006, Gennaioli and Rainer 2007). The ethnic institutional structures become salient as most African states characterised by limited state capacity are unable to provide public goods, collect taxes, and enforce contracts (Acemoglu 2005, Besley and Persson 2009). For example tribal chiefs, rather than national judges tend to resolve disputes, especially in the African country side. Moreover, local chiefs rather than state bureaucrats collect taxes, enforce property rights, and provide public goods.
In our recent paper (Michalopoulos and Papaioannou 2010), we examine the effect on comparative development of national contemporary institutional structures and ethnicity-specific pre-colonial institutional traits. To advance on the causality front we develop a methodological approach that combines anthropological data on the spatial distribution of ethnicities before colonisation, historical information on some key ethnic institutional/cultural traits, and contemporary light density image data from satellites as a proxy of regional development. Figure 1a illustrates the historical homelands of the major ethnicities spanning the African landscape in the eve of colonisation. Figure 1b plots satellite density across each tribal area. One observation that becomes immediately obvious is the significant variation in the degree of luminosity across regions.
Contemporary national institutions and regional development
To identify the one-way effect of contemporary national institutions on regional development we exploit in a quasi-experimental econometric setting the so-called “scramble of Africa” in the late 19th century. The partitioning of Africa started with the Berlin Conference of 1884-1885 that established the principles on how Europeans would split Africa into spheres of influence, protectorates, free-trade areas, and colonies and was completed by the beginning of the 20th century. Crucially the borders that Europeans draw in the map in Berlin, London, Brussels, and Paris at a time when they had not even settled Africa, endured after independence in the 1960s. As a result of this arbitrary border drawing, hundreds of ethnic groups were partitioned into more than one state, thus subjecting identical cultures to different country-level formal institutions.
We make use of this historical accident. We compare economic performance as proxied by satellite light density at night in regions belonging to the historical homeland of the same ethnic group, but fall into two (or more) countries that have national institutions of different quality.
Our approach required identifying partitioned ethnicities. To identify split groups we combine George Peter Murdock’s (1959) Ethnolingusitic Map that portrays the spatial distribution of 835 ethnicities before colonisation (Figure 1) with the Digital Chart of the World that delineates contemporary political boundaries. Our procedure identifies most ethnicities that case studies and anecdotal evidence suggest that have been partitioned by the political boundaries (see Figure 2). For example, the Maasai have been partitioned between Kenya and Tanzania (shares 62% and 38% respectively), the Anyi between Ghana and the Ivory Coast (shares 58% and 42%), and the Chewa between Mozambique (50%), Malawi (34%), and Zimbabwe (16%).
Our econometric (regression discontinuity) approach compares economic performance (as reflected in luminosity density) in the historical homeland of the Anyi in Ghana with that in the adjacent region in the Ivory Coast where the Anyi also reside. We also do the study for the Maasai between Kenya and Tanzania. This “matching” type approach that focuses on adjacent by the border regions belonging to the historical homeland of the same ethnicity allows us to account for hard-to-measure cultural features (related to trust, religion, societal habits) that correlate significantly with both economic wellbeing and sound institutions. Moreover by focusing on nearby areas we account effectively for geography, the disease environment, and other ecological features; this is fundamental as many influential scholars argue that both institutional and economic development are driven by geographic-ecological difference (Sachs 2005).
In line with casual empiricism we find that regional development is higher in ethnic areas with better functioning national institutions – as reflected in a composite rule of law index that captures effectiveness of the judiciary and the quality of property rights protection as well as the control of corruption.1 Yet our within-ethnicity results show that there is no systematic relationship between country-wide differences in institutions and tribal economic performance (see Figures 3a, 3b). This result casts doubt on the causal interpretation of the cross-country positive correlation between countrywide institutional quality and economic development in Africa.
Figure 3a. Does corruption matter?
Figure 3b. Does rule of law matter?
Pre-colonial ethnic institutions and regional development
We then examine the effect of deep historical ethnicity-specific institutional characteristics on African development utilising two measures from the pioneering work of George Peter Murdock (1967), who through extensive field work and research, has produced a database of quantitative measures of cultural, economic, and institutional characteristics of African ethnicities at the time of colonisation. Figures 4a and 4b portray the vast heterogeneity in political centralisation and ethnic class stratification across African ethnicities. The political centralisation index ranges from 0 to 4, reflecting the number of jurisdictional hierarchies above the local community level for each society. Previous work suggests that by increasing the accountability of local chiefs, politically centralised societies can achieve better economic outcomes (Gannaioli and Rainer 2007). Examples of large ethnicities without any level of political organisation above the village level include the Bura and the Lango in Uganda. Examples of tribes belonging to small chiefdoms are the Mende in Sierra Leone and the Ibo of Nigeria. The Mbundu in Angola and the Zerma in Niger are examples of societies that are part of large paramount chiefdoms, while the Yoruba in Nigeria, the Mossi in Burkina Faso and the Kafa in Ethiopia are parts of large states. As with the jurisdictional hierarchy measure, there was significant heterogeneity in class stratification across Africa at the time of colonisation. Highly stratified societies (with complex distinctions) include the Yoruba and the Nupe in Nigeria, while examples of ethnicities without wealth or elite distinctions are the Maasai in Kenya and Tanzania, the Chewa in Malawi, and the Songo in Angola.
Our econometric analysis establishes that political centralisation and statehood experience before the advent of European colonisers correlates significantly with contemporary development. The descriptive evidence reported in Figure 5a shows that light density is significantly higher in areas where politically centralised societies reside. The same pattern applies when we examine the correlation between luminosity and class stratification (Figure 5b), a societal trait that has been linked to property rights protection (e.g. Rudmin 1995; Acemoglu et al. 2008).
In our econometric analysis we estimate various empirical specifications to account for possible confounding factors producing the correlation patterns shown in Figures 5a and 5b.2 The strong positive correlation between ethnic pre-colonial institutions (and class stratification) and regional development prevails when we control at a very fine level for urbanisation and a rich set of geographic variables (capturing the disease environment, land suitability for agriculture, elevation, etc). Most importantly pre-colonial ethnic institutions correlate significantly with regional development even when we solely utilise the within country variation, so as to account for unobserved country-level attributes, reflecting national policies and institutions, foreign aid, etc. Figures 6a-b summarise our key results on the effect of political centralisation and class stratification. Each bar corresponds to a different econometric model that
- does not control for other factors (in yellow),
- nets out the effect of urbanisation (in orange),
- also accounts for geography (in light red),
- includes country fixed-effects to account for all national features (in dark red),
- controls for urbanisation, geographical/ecological features and national characteristics (in dark blue).
The bars in Figure 6a reflect how much higher regional development is in ethnic areas of politically centralised societies (where the jurisdictional hierarchy index equals 2, 3, and 4) as compared to regions where stateless ethnic groups reside (the jurisdictional hierarchy index equals 0 or 1). Likewise the bars in Figure 6b show how much higher development is in areas where stratified societies reside as compared to regions populated by egalitarian/communal societies.
In all specifications the estimated effects is significant at standard confidence levels. Our preferred empirical models (the one where we control for urbanisation, geographical features, and all nation-wide factors) imply that regional development is approximately 5%-6% higher in areas where socially complex and politically centralised societies reside.
We also find that the positive effect of ethnic-specific institutions on regional development persists and if anything is stronger across ethnic groups partitioned by the national boundaries. This result suggests that along these territories characterised by the negligible penetration of countrywide policies it is the tribal institutions, determined well before the colonial era, that shape contemporary development.
Our analysis emphasises the importance of ethnic rather than national institutional structures in explaining African development. The key message of our empirical work is thus to shift the attention from national factors to ethnic-specific institutional, cultural, and organisational norms. While our finding that national institutions play no role in explaining differences in regional development looks at first glance paradoxical, it is in line with a multitude of African case studies and anecdotal evidence that stress the limited penetration of national policies across the continent (Dowden 2008). Moreover our result on the strong effect of ethnic rules is in line with an influential conjecture among African historians that deeply rooted ethnic societal and institutional features still govern economic activity in many parts of Africa (Herbst 2001).
Besides these findings, our codification of partitioned ethnic groups and the combination of high resolution regional data on development (such as satellite light density at night) with historical measures on culture and institutions can provide a platform for subsequent research to address unresolved topics on the fundamental determinants of comparative economic development.
Acemoglu, Daron (2005), “Politics and Economics in Weak and Strong States”, Journal of Monetary Economics, 52(6):1199-1226.
Acemoglu, Daron, Simon Johnson, and James A Robinson (2005), “Institutions as a Fundamental Cause of Long-Run Growth” in Philippe Aghion and Steven Durlauf (eds.), The Handbook of Economic Growth, Elsevier.
Acemoglu, Daron, Maria Angelica Bautista, Pablo Querubin, and James A Robinson (2008), “Economic and Political Inequality in Development: The Case of Cundinamarca, Colombia”, in E Helpman (ed), Institutions and Economic Performance, Harvard University Press.
Alesina, Alberto, William Easterly, and Janina Matuszeski (2010), “Artificial States”, Journal of the European Economic Association.
Besley, Tim and Thorsten Persson (2009), “The Origins of State Capacity: Property Rights, Taxation and Politics”, American Economic Review, 99:1218-1244.
Dowden, Richard (2008), Africa: Altered States, Ordinary Miracles, Portobello Books Ltd.
Englebert, Pierre,.Stacey Tarango, and Matthew Carter (2002), “Dismemberment and Suffocation: A Contribution to the Debate on African Boundaries”, Comparative Political Studies, 35(10):1093-1118.
Gennaioli, Nicola and Ilia Rainer (2007), “The Modern Impact of Precolonial Centralization in Africa”, Journal of Economic Growth, 12(3):185-234.
Herbst, Jeffery (2001), State and Power in Africa, Princeton University Press.
Michalopoulos, Stelios and Elias Papaioannou (2010), “Divide and Rule or the Rule of the Divided? Evidence from Africa”, CEPR Discussion Paper 8088.
Nunn, Nathan (2008), “The Long Term Effects of Africa’s Slave Trades”, Quarterly Journal of Economics, 123(1):139-176.
Rudmin, Floyd Webster (1995), “Cross-Cultural Correlates of the Ownership of Private Property: Two Samples of Murdock’s Data”, Journal of Socio-Economics, 24(1):345-373.
Sachs, Jeffrey (2005), The End of Poverty, Penguin.
1 We obtain similar results when we use other proxy measures of national institutional quality, such as bureaucratic efficiency, executive constraints, legal formalism, etc.
2 We also conducted various robustness checks to account for measurement error in the two proxy measures of ethnic-specific institutional capacity and other concerns.