Chapter 1 provides a brief overview of the recent literature in economic history and long-run development, and summarizes the main findings of the three essays presented in this dissertation.
In Chapter 2, I examine the subject of villagization in Tanzania, a major episode of development planning in post-independence Tanzania. I revisit this period of Tanzania’s economic history, focusing on the legacy of developmental villages (vijiji uya maendeleo) introduced in mainland Tanzania over the period 1974-1982. Combining historical data on Tanzania from the 1970s with data from population censuses and recent national household surveys, I investigate whether variation in the intensity of the governments villagization program explains within-region variation in social and economic outcomes today. I document that, in the short-run, developmental villages led to an increase in various educational outcomes, such as primary school completion rates, literacy rates, and total years of schooling. Today, districts which experienced a high share of developmental villages have greater availability of some public goods and citizens report higher rates of participation in community activities, but there is worse perception of corruption among government officials and greater rejection of one-party rule. Per capita household consumption is also significantly lower in districts with historically high levels of the treatment measure. To address potential endogeneity in village formation, I report instrumental variable results based on variation in ethnolinguistic fragmentation and the occurrence of droughts in the 1970s which facilitated the resettlement of peasants into villages. I conclude by providing some preliminary evidence on the lack of economic diversification as well as political alignment to the TANU/CCM party as possible channels which explain the legacy of the villagization experiment.
In Chapter 3, I turn to the subject of disease eradication, and examine the impact of the successful control of a highly infectious tropical disease, yaws, in Ghana over the period 1956-1963. The availability of cheap, mass-produced penicillin following World War II resulted in a mass treatment campaign by WHO/UNICEF aimed at controlling the prevalence of yaws and other bacterial infections. I examine the effect of this penicillin campaign in which over 70 percent of the estimated Ghanaian population received a single dose of an intramuscular penicillin injection. Data collected by the WHO/UNICEF program before and after the campaign indicates that penicillin-based treatment resulted in an immediate reduction in the prevalence of infectious yaws among the Ghanaian population. Using a microsample from the 2000 Ghanaian census, I estimate a difference-in-difference model exploiting spatial variation in pre-treatment prevalence of yaws infections and variation in exposure due to the timing of the penicillin campaigns. My results indicate that, following the penicillin campaigns, cohorts born in districts with higher initial yaws prevalence achieved higher education outcomes than prior generations when compared with cohorts from districts with lower yaws prevalence. The results are particularly robust for the female subsample, where I observe increases in educational attainment for cohorts born just prior to the penicillin campaigns.
In Chapter 4, I study the development of political partisanship, examining the plausibly random spread of the cocoa swollen shoot disease in the Gold Coast/Ghana in the 1940s. In 1948, the Watson Commission which investigated riots in colonial Ghana sparked by the cocoa swollen shoot pest noted the political motivations of the disturbances. In this chapter, I utilize novel data on cocoa farm acreages and the spatial variation in the spread of the swollen shoot virus to investigate the impact of the pest on the development of local political movements. Based on responses from the Afrobarometer surveys, I find that today, individuals in districts which historically experienced a high intensity of the disease pest report stronger anti-government opinions, and are more likely to attribute success in life to individual effort than government support. I trace the historical roots of these political views by examining electoral results from the 1956 Legislative Elections in colonial Ghana. Conditional on region fixed effects, and various pre-epidemic district controls, I observe that more adversely affected districts were more likely to vote against the new centerleft (Nkrumahist) government. By 2000, with multiparty democracy, these areas still vote against the center-left (Nkrumahist) party. This partisan opposition has an impact on the allocation of resources today. Using an instrumental variable strategy, I examine the impact of government opposition on local government transfers received in various districts, with the historic intensity of the pest shock as an instrument. I examine possible violations to the exclusion restrictions of the 2SLS strategy by ruling out the impact of the cocoa swollen shoot disease on other economic and social outcomes. Based on the approach developed by Conley, Hansen and Rossi (2012), I also document that the 2SLS results remain robust to moderate forms of violations to the exclusion restriction assumptions.