Explaining ethno-regional favouritism in Sub-Saharan Africa

A burgeoning literature on ethno-regional favouritism in Sub-Saharan Africa has found that Presidents favour their co-ethnic kin in the provision of public and private goods. However, this scholarship has largely remained empirically narrow in focus, inasmuch as it preponderantly examines only one outcome and/or country at a time and can be contrasted with a separate set of literature which finds a null or even negative relationship between co-ethnicity and goods provision. As such we conduct the largest examination to date of ethno-regional favouritism in Sub-Saharan Africa using data from the Afrobarometer and DHS across both public and private goods and at both the individual and district level. We confirm the positive effects of individual-level co-ethnicity on a variety of outcomes, but also find that these benefits only accrue to the few co-ethnics living in non-coethnic areas and decline as the district-level proportion of co-ethnics increases. The positive effects of individual-level co-ethnicity are weaker for objective outcomes like access to infrastructure, asset ownership and employment but are stronger for subjective measures such as self-assessed living conditions and the quality of government services. We also find that the positive effects of co-ethnicity do not decline with the proportion of local co-ethnics for subjective perceptions of presidential and ruling party performance. This relationship does not hold, however, for perceptions of other non-political institutions like the courts or police, or for local governments. These results are consistent with the argument that co-ethnics derive non-material "psychic goods" from having a co-ethnic in power, rather than the standard "quid-pro-quo" theory common in the literature, and thus complicate the idea that ethnic favouritism in the provision of public and private goods is widespread in contemporary Africa. We supplement our quantitative findings with anecdotal evidence from Nigeria which supports our argument.