Essays on ethnic diversity and political instability in sub-Saharan Africa

This dissertation investigates the relationship between ethnic diversity and different forms of political instability in sub-Saharan Africa. The first essay, written with Kimuli Kasara, is a cross-national statistical study of the effect of the sizes of ethnic groups and their capacities for revolt on political instability, defined broadly as violent, irregular attempts to take over national power. We construct and test several hypotheses from the observation that only small numbers of people are needed to challenge the state and find partial support for our view. In a transitional model analysis on data covering independence through 2004, we find evidence that the effect of state strength and ethnic groups' capacity to challenge the state is conditional upon a country's recent history of political instability.; The second and third essays develop the idea that resources that can win political support from potential opponents may also strengthen them. In weak states where many political actors have or could build the capacity to use violence in politics, dispensing patronage may be dangerous.; The second essay presents two versions of a formal model of this dynamic to investigate how ethnic diversity affects the likelihood of a revolt or a coup d'etat against the president. The models assume, as in the first essay, that the sizes of ethnic groups do not affect the probability that a group will he successful should it attempt to overthrow the president. In the second version of the model, there is an election if there is no revolt, and larger ethnic groups have an advantage in these elections. If the ethnic group that does not hold the presidency has a low baseline capacity for revolt, and is inefficient in converting patronage into greater capacity for revolt, political instability is more likely when the president's ethnic group is numerically dominant. However, if this condition does not hold, then the probability of revolt is weakly increasing and then weakly decreasing in the size of the president's ethnic group. This essay presents some preliminary analyses on African data and discusses future directions for empirical testing and further theoretical development.; The third essay considers how political parties are organized when party leaders face this dilemma in dealing with local politicians. It examines the consequences of this type of party organization for campaigning and political thuggery in three states with differing ethnic demographics in Nigeria, a country which only returned to civilian rule in 1999. Winning elections in Nigeria requires a network of local politicians whose influence ''on the ground" can win the support of voters and elections officials. But these local politicians cannot be trusted with the party leader's resources in advance of the elections, as they may simply abscond with the funds, support another party, or even use the funds to cultivate personal thugs.; To address this problem, party leaders organize their parties as rank-order tournaments, in which local politicians use their own resources and compete against one another for political appointments to be made by the party leader after the election. The party leader tries to set incentives so that local politicians direct their resources towards turning out supporters for the party, and this is the outcome in ethnically homogeneous states with multiple parties. These incentives break down, leading to thuggery among local politicians and against the party leader, however, when there is insufficient inter-party competition at the local level and wide disparities in the competitiveness of the party in different local areas due to the geographical concentration of ethnic groups. This essay presents a formal model of tournament parties and presents supporting data collected during twelve months of field work in Ekiti, Os&dotbelow;un and Kwara states in Nigeria around the 2003 general elections.