UC Berkeley

While sub-Saharan Africa makes up only one-tenth of world population, it contains two-thirds of all the HIV infections worldwide. This dissertation examines individual behavior in the context of the HIV/AIDS epidemic in sub-Saharan Africa.

Chapter 2 examines the effects of HIV testing on risky sexual behavior. Using data from a study that randomly assigns offers of HIV testing in two urban centers in East Africa I examine the effects of testing, taking into account people's beliefs about their HIV status prior to testing. I find large behavioral responses to HIV tests when tests provide new information to individuals. Individuals surprised by an HIV-positive test increase their risky sexual behavior, while individuals surprised by an HIV-negative test decrease their risky sexual behavior. When HIV tests agree with a person's belief of HIV status there is no change in sexual behavior. Using these estimates, I simulate the effects of testing on new HIV infections. Using the distribution of beliefs of HIV infection and prevalence from the study, I find the overall number of HIV infections increases by 25% when people are tested compared to when they are unaware of their status - an unintended consequence of testing.

Chapter 3 explores the effects of income shocks on behavior that might lead to increases in the risk of HIV transmission. Such behavior includes increasing the frequency of, or risks taken during, transactional sex, or expanding one's sexual network to acquire informal insurance against future shocks. We present here evidence of the impacts of this phenomenon on a widespread scale in sub-Saharan Africa. Lacking modern irrigation, substandard rainfall in Africa reduces crop yields, potentially inducing economic hardship, especially in rural areas. We find that each local shock of this kind over the preceding 10 years predicts an increase in HIV infections in rural women of up to 13%, depending on the existing prevalence. Further, the evidence suggests that the effects are concentrated among the most vulnerable women -- those with low levels of wealth and education.

Chapter 4 examines the effects of education on rural to urban migration in an HIV epicenter. The Kangera region in northwestern Tanzania is believed to be the epicenter of the AIDS epidemic in East Africa with HIV prevalence estimated at 24% in the regional capital during the early 1990's. Using the Kangera Health and Development Survey (KHDS), I observe both educational attainment and migration decisions. To control for the endogeneity of education, I exploit a change in Tanzania's national policy for secondary education to identify exogenous variation in schooling. This variation creates an instrumental variable for education. Overall, there is a positive and statistically significant correlation between educational attainment and the three measures of migration. However, once education is instrumented, the relationship between education and migration is inconclusive.