Probability Limits: Are Subjective Assessments Adequately Accurate?
The Health and Retirement Study asks respondents their subjective probabilities about 12 future events. An individual's responses contain a common component that is unrelated to the true probability of the event in question. Use of the entire set of an individual's responses to control for this unobserved individual heterogeneity can improve the information content in responses regarding intergenerational transfer and labor force participation plans. Although there is little overall gain from renormalization, in samples where the respondent may not fully have understood the question adjusting the responses for heterogeneity leads to an improved ability to predict outcomes in later waves.