June 1, 2006
This article considers dynamic models of smoking under uncertainty, wherein individuals learn about the associated risks both through experimentation and observation. We use smokers' changes in self-reported health in long-run household panel data as an individualized measure of information about how dangerous smoking is for them. We find that own past health changes while smoking are often positively correlated with current cigarette consumption, implying learning. However, other household members' health changes have only a weak effect. We emphasize individual heterogeneity, using both fixed effects in the analysis of cigarette consumption and age and sex differences in how individuals react to health changes. We conclude that smokers do indeed react to personalized health information (but not necessarily from generalized information) and modify their behavior accordingly. The way in which they react differs sharply by sex and by age, suggesting that aggregate correlations may be misleading.
Substance Use and Misuse
Volume: 41 (4):427-451
Albert Sloman Library Periodicals *restricted to Univ. Essex registered users*