The Effect of HIV Infection Risk Beliefs on Risky Sexual Behaviors: Scared Straight or Scared to Death?

Economists typically assume that risk compensation is uniformly self-protective – that people become more careful as the health risks of their actions increase. However, risk-seeking, or fatalistic, responses can also be rational: increased risks can lead people to take fewer precautions. I extend the typical model of risk compensation to show that fatalism is a rational response to sufficiently high risks if people do not have perfect control over all possible exposures, and if the condition in question is irreversible. This result holds even for people who do not understand how to add up probabilities. I test this model’s implications by randomizing the provision of information on HIV transmission risks to people in Malawi, a country with a severe HIV epidemic where there is qualitative evidence of fatalistic responses to the virus. Average risk responses are self-protective and statistically significant, but small in magnitude: the mean risk elasticity of sexual behavior is roughly -0.6. To test the model of rational fatalism, I develop a method of decomposing 2SLS estimates of the risk elasticity of sexual behavior by baseline risk beliefs. Consistent with the predictions of my theoretical framework, I find that this elasticity varies sharply by baseline risk beliefs: the risk elasticity varies from -2.3 for the lowest initial beliefs to 2.9 for the highest initial beliefs. 13.8% of the population has a positive elasticity, suggesting they are fatalistic. ∗University of Michigan Department of Economics and Population Studies Center. Email [email protected]. I am grateful to Rebecca Thornton, Jeff Smith, John DiNardo, David Lam, Matias Cattaneo, Bob Willis, Mel Stephens, John Bound, Miles Kimball, Susan Watkins, Pinar Keskin, Daniel Bennett, Nate Young, Sanyu Mojola, Ophira Vishkin, Evan Herrnstadt, Joe Golden, Audrey Dorélien, Laura Derksen, Eric Chyn, Olga Malkova, Justin Ladner, and seminar participants at the University of Michigan, PAA, MIEDC, ABCDE, and NEUDC for their valuable feedback. This project would not have been possible without the excellent fieldwork supervision efforts of Anderson Moyo, Synab Njerenga, and Christopher Nyirenda. Data collection on this project was supported by internal UM grants from the Center for Global Health, the Population Studies Center, the Weinberg International Travel Fund, the Center for Education of Women, and the Michigan Institute for Teaching and Research in Economics as well as the Russell Sage Foundation’s Small Grants in Behavioral Economics program. I acknowledge fellowship support from the Population Studies Center and the Rackham Graduate School. All errors and omissions are my own.