John Doces and Erik Heinemann (Bucknell University)
In Africa, a common theme in development is that authoritarianism has been detrimental to development. In particular, arguments about the nature of this relationship focus on the role of the African “Big Man” and the effect of patronage politics viewing the situation as one in which people connected to the ruler benefit most from his rule. Most considerations of this idea, however, focus on country-level indicators, but this does not precisely test the underlying argument because the expected level of variation in development outcomes is within the country not necessarily across countries. To test this argument we thus focus within Uganda and how “Big Man” rule has affected development there. We do this using a sub-county spatial analysis of Uganda employing GIS mapping techniques and a regression analysis to test if there is an effect of “Big Man” rule and sub-county development. To measure the influence of the Big Man, here Yoweri Museveni, we calculate each sub-county’s distance to his birthplace as well as distance to Entebbe, Uganda’s capital city. We expect both to be inversely associated with development at the sub-county level. To measure development we use indicators of sanitation including the percentage of households with access to a latrine and access to soap for hand-washing. Our spatial maps show a clear association between distance and development indicating that the further a sub-county is from where Museveni was both there are lower levels of development. Moreover, our regression analysis shows that controlling for a number of other variables—e.g., poverty rate, poverty density, population density, urban sub-county, and total households—that the effect of distance from both Museveni’s hometown and distance from Entebbe both are statistically significant and inversely associated with our development indicators. Specifically, a one standard deviation increase in distance from Museveni’s hometown is associated with a fall in latrine coverage by half of a standard deviation or about 32% less coverage. Alone the two measures of distance explain roughly 45% of the variation in development outcomes and the full models explain almost 60% of the variation.