Survey measures versus incentivized measures of risk preferences: Evidence from sex workers' risky sexual transactions.

Survey measures of risk attitudes are primarily used in the health literature, although incentivized measures of risk preferences are being increasingly used in other fields. We exploit the unique setting of commercial female sex workers in Bangladesh to investigate whether incentivized measures of risk preferences, or non-incentivized survey measures of risk preferences, best identify the risky commercial sex decisions that they make. The study uses survey data collected during February-April 2016, and October-November 2016 from eight brothels in Bangladesh. Wave 1 includes 1,332 female sex workers, Wave 2 includes 1,185 female sex workers. Our findings suggest that researchers can reliably use survey measures to elicit risk preferences on health.