Evaluation of Subjective Probability Distributions in the HRS

In the Health and Retirement Survey respondents were asked about the chances they would live to 75 or to 85, and the chances they would work after age 62 or 65. We analyze the responses to determine if they behave like probabilities, if their averages are close to average probabilities in the population, and if they have correlations with other variables that are similar to correlations with actual outcomes. We find that generally they do behave like probabilities and they do aggregate. Most remarkable, however, is that they covary with other variables in the same way actual outcomes vary with the variables. For example, smokers give lower probabilities of living to 75 than nonsmokers. We conclude that these measures of subjective probabilities have great potential use in models of intertemporal decision making under uncertainty.