Subjective beliefs, health, and health behaviors

Abstract

This article reviews economic studies based on data from high income countries published from 2007 to early 2024 to address three questions: (1) How accurate are subjective beliefs, mainly measured by subjective probabilities, compared to their objective counterparts? Objective evidence comes from another source (e.g., life table, empirical study, expert opinion), or subsequent realizations of beliefs elicited at baseline. (2) How are subjective beliefs determined? (3) Do subjective beliefs affect health behaviors? Several domains are included: survival, and health behaviors—smoking, alcohol consumption and impaired driving, preventive care, diet, and COVID-19 precautions. Results on a single domain, (e.g., survival), do not generalize to, e.g., COVID-19 results. Subjective probabilities embody private information (e.g., self-assessed health, parent longevity). However, individuals seem insufficiently informed about population-level probabilities. There is no systematic overestimation or underestimation of objective probabilities. Several determinants of beliefs are identified (demographic characteristics, education, cognition, current self-assessed health, health histories), but evidence on underlying mechanisms is lacking, how determinants, (e.g., education), affect beliefs. Subjective beliefs, even with substantial noise, often affect health behaviors. Given prior evidence that beliefs are influenced by health shocks, this article reviews research on effects of health shocks on health behaviors. A major health shock to an individual—a new diagnosis (e.g., diabetes) or a serious adverse health event (e.g., heart attack), by changing subjective probabilities leads to some healthier behaviors, however, sometimes only temporarily. Behaviors may also be influenced by utility loss following a health shock, e.g., learning about pecuniary and non-pecuniary costs of hospitalization.