Antiretroviral drug access and behavior change

The rapid expansion in access to antiretroviral (ARV) drugs in sub-Saharan Africa from under 10,000 in 2000 to more than 8 million in 2011 has been enormous. Anticipating the impact of this expansion and future policy options requires identification of the impacts of access on individual behaviors that determine the course of the epidemic. This paper combines geocoded information about the timing of introduction of ARVs in each health facility in Kenya with two repeated cross sections of geocoded population surveys to estimate the impact of ARV access on risky sexual behavior. Using difference in differences estimates, I find evidence of a relative increase in risky behavior reflected in pregnancy rates among young people in areas with ARV access. These results are consistent across different definitions of treatment and different sets of fixed effects. Using a simple model, I show that the reduction in transmission from ARVs can outweigh the increase in risk-taking in determining new infections. JEL Classification: I12, I18, J13, O15 Health Health Outcomes ∗Financial support for data collection was received from the 2011 Weiss Family Fellowship through the Center for Evaluation and Global Action. The following individuals provided valuable help and support in the collection and accessing of the data: Evan Klaus, Denis Kimera, Reinier Mountbatten, Sabin Nsanzimana, Jean-Paul Uwizihiwe, Jean de Dieu Bizimana, Ruton Hinda, Patrick Mwangi, Ibrahim Mohamed, Susan Njogo, Megumi Gordon and Michael Laverty. I am particularly grateful to Edward Miguel, Ernesto Dal Bo, Frederico Finan, and Justin McCrary for extremely helpful advice with this project.