Research on African favouritism suffers from endogeneity issues
This is an important finding – not only because it raises doubt about some of the quantitative evidence of ethnic favouritism in Africa, but also because it urges us to think about the policy interests and development models that shaped the thinking of early independence leaders in Africa. Many independence leaders sought to accelerate the late colonial-era development policies (particularly in education), rather than radically alter them, possibly because they came from regions that had been comparatively well served by such policies.
In related work, Elliott Green and I showed other problems with these ethnic favouritism models, and we found that the supposed evidence on ethnic favouritism in Kenyan education is far from robust (see: Simson and Green, 2020, ‘Ethnic favouritism in Kenyan education reconsidered’, Journal of Modern African Studies). This is in part for the same reasons that you discuss – that Kenya’s first President, Jomo Kenyatta, came from the Kikuyu ethnic group, which had the highest educational attainment during the colonial era. By measuring favouritism on the basis of years of primary schooling, which has a hard upper limit, the rate at which Kikuyu primary school attainment grows will slow as this group starts approaching universal attainment. Consequently, the primary schooling gap between the Kikuyu and other ethnic groups looks larger under Kenyatta’s rule than under that of his (non-Kikuyu) successor, and vice versa for the non-Kikuyu groups. These convergence effects, coupled with the fact that the first ruler after independence was often from an educationally-advantaged region of the country, make it quite complicated to model favouritism.
There are also other problems with these models. For example, much of the school construction in Kenya in the 1960s and 1970s was financed and undertaken by local communities rather than the central government, so the rate at which primary schooling expanded was in large part demand-driven. You could argue that President Kenyatta favoured this model because it served the interests of the Kikuyu (who were better placed to make schooling investments), but this is a rather far-fetched notion of ethnic favouritism, and one can make a strong developmental case for allowing such self-financed, demand-driven service expansion, even if it is regressive. In fact, I think the ethnic favouritism literature on Africa would benefit from a more traditional study of the distributional consequences of different policy regimes, and how the background of politicians, and the constituencies they represent, shape these policy preferences.
Thanks for sharing, and will definitely read the longer paper.
What would be interesting would be the factors that led to colonial social investment within a particular area i.e. are there other pre-colonial institutional factors which led to a larger concentration of missionaries within an area, which then led to more educational institutions, which then led to more post-colonial leaders.