Essays on social networks in developing countries

Social networks can be vital in developing countries where institutions and markets are missing or incomplete. Social networks can supplement markets that are not suitable for people’s economic needs by sharing resources, shaping beliefs and transmitting information. This thesis studies social networks in developing countries, which is important to find effective approaches to deal with the problems. The first chapter examines the peer effects of religious groups on learning one’s own HIV status. This analysis is interesting because religion substantially affects the lives of people in sub-Saharan Africa and there could be peer effects within religious groups which drive behaviour including stigma. It is important to understand these in order to design policies to overcome stigma. I use exogenous variation from randomised incentives to learn test results in rural Malawi (Thornton, 2008) to address difficulties in disentangling peer effects from other effects. I find that the proportion of religious group learning their HIV results increases the probability of a group member’s finding out their HIV test results. The second chapter studies peer effects on subjective expectations of HIV. Subjective expectations significantly influence health behaviours and play a crucial role in health outcomes by inducing take-up of preventive health behaviours and avoidance of risky behaviours. I use data from Delavande and Kohler (2009) which collected subjective expectation on one’s own HIV status by easy visual methods in rural Malawi. The results show significant and positive peer effects on one’s own subjective likelihood of HIV infection. The third chapter explores the effect of rainfall shock on within-village extended family networks. Within-village extended family networks, and the interactions within these networks, play an important role in risk and resource sharing. The influences of rainfall shocks – measured as absolute deviation from long-term average – on the structure of within-village extended family networks in rural Mexico are examined. In villages where reduction in incomes is implied due to rainfall shocks, extended family networks have smaller degree and size. However, these negative effects are countervailed by cash transfers. Possible explanations for these results are the decrease of females leaving their households for marriage in control group and reduced work migration in treatment group with cash transfers. The findings in this thesis emphasise the importance of social networks in developing countries in terms of behaviour, belief and risk-sharing.