The regional composition of a government affects conflict, clientelism and public goods provision in developing countries, many of which are former colonies. But what explains how power is distributed across regions to begin with? Given the strategic nature of cabinet formation, extant explanations focus on bargaining---leaders allocate portfolios strategically to minimize unrest---but fail to consider long-term factors. Using a sample of 16 former British and French African colonies, I find that some colonial districts were represented in post-colonial governments much more than others even adjusting for population. By combining historical records and geospatial data, I show that this regional political inequality derives from colonial investments in public (missionary) education in French (British) colonies but not from other colonial investments, levels of development during colonialism or pre-colonial factors. I argue that post-colonial ministers are a byproduct of a civil service recruitment practice among European administrators focused on natives' literacy rather than ethnicity. Thus, regional political inequality after the end of foreign rule has a structural human capital component which may mediate the relationship between colonialism and current political and economic development.
Recent literature has documented the economic and political consequences of colonialism, but we know less about the origins of colonial investments. I present evidence that they were very unequally distributed within 16 French and British African colonies, even when adjusted for population. How did colonial powers allocate their investments? Economic history emphasizes the role of natural resources while recent literature in political economy argues that settlers explain the origin of colonial institutions. I argue that observable geographic features---locational fundamentals---led some locations to become centers of pre-colonial trade, which in turn increased colonial settlement and investments not only in infrastructure but also in health and education. Disparities did not diminish during the colonial period because those locations became centers of colonial economic activity and benefited from complementarities between investments, consistent with a logic of increasing returns.
Culture is a central but elusive concept in the social sciences, and so are its effects. We take advantage of a natural setting-the oldest university in East Africa-where students are randomly assigned to live in halls of residence that have maintained distinct "student cultures" since the 1970s. A broad student consensus at the university characterizes certain halls as sociable and activist, and others as academically-minded and respectful. Using an original survey of current students and behavioral games, we find that the hall cultures on campus do not influence students' academic performance, social habits or ideology. However, they affect same-hall peer characterization, time preferences, identity, and interpersonal trust and generosity. A second survey, of alumni, suggests that some of these cultural effects are enduring, and influence activism-while on campus and today-as well as the choice of marriage partners.
This paper presents three related ndings on regional decentralization. We use an original dataset collected in Uganda to establish, for the first time in a developing country context, that individuals have meaningful preferences over the degree of regional decentralization they desire, ranging from centralism to secessionism. Second, multilevel models suggest that a small share of this variation is explained at the district and ethnic group levels. The preference for regional decentralization monotonically increases with a group or district's average ethnic attachment. However, the relationship with a group or district's income is U-shaped: both the richest and the poorest groups desire more regionalism, reconciling interest and identity-based explanations. Finally, we show that higher individual ethnic attachment causes an increase in regionalist preferences using fixed eects and a new matching method for general treatment regimes.