Dynamic optimization and conformity in health behavior and life enjoyment over the life cycle.

ABSTRACT

This article examines individual and social influences on investments in health and enjoyment from immediate consumption. Our lab experiment mimics the problem of health investment over a lifetime (Grossman, 1972a,b). Incentives to find the appropriate expenditures on life enjoyment and health are given by making in each period come period a function of previous health investments. In order to model social effects in the experiment, we randomly assigned individuals to chat/observation groups. Groups were permitted to freely chat between repeated lifetimes. Two treatments were employed In the Independent-rewards treatment, an individual's rewards from investments in life enjoyment depend only on his choice and in the Interdependent-rewards treatment; rewards not only depend on an individual's choices but also on their similarity to the choices of the others in their group, generating a premium on conformity. The principal hypothesis is that gains from conformity increase variance in health behavior among groups and can lead to suboptimal performance. We tested three predictions and each was supported by the data the Interdependent-rewards treatment (1) decreased within-group variance, (2) increased between-group variance, and (3) increased the likelihood of behavior far from the optimum with respect to the dynamic problem. We also test and find support for a series of subsidiary hypotheses. We found (4) Subjects engaged in helpful chat in both treatments; (5) there was significant heterogeneity among both subjects and groups in chat frequencies; and (6) chat was most common early in the experiment, and (7) the interdependent rewards treatment increased strategic chat frequency. Incentives for conformity appear to promote prosocial behavior, but also increase variance among groups, leading to convergence on suboptimal strategies for some groups. We discuss these results in light of the growing literature focusing on social networks and health outcomes.