Mortality Risk Perceptions: A Bayesian Reassessment

Abstract

This paper uses a Bayesian learning model to assess the respective influence of different risk measurements on mortality risk perceptions. People form risk beliefs using several sources of information, including the actual population mean death risk level, the discounted lost life expectancy, and the age-specific hazard rate considered by Benjamin and Dougan (1997). The appropriate criterion for judging the validity of risk perceptions is not the perfect information case, but rather whether people form their risk beliefs in a rational manner given a world of costly and limited risk information. Although the statistical results support the overall conclusion that the learning process is rational, the character of the learning process differs depending on the risk level. Risk-related variables are much better predictors of larger risks than of small risks, which reflects the role of information costs and the benefits of learning about larger risks.