Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Because of the severe long-term health consequences of smoking, there has been much interest in economics, health sciences, sociology, and other disciplines in documenting the determinants of adolescent smoking and suggesting policy interventions to reduce initiation and increase cessation. As there is mixed evidence that cigarette price is an important determinant of adolescent smoking decisions (Chaloupka and Wechsler 1997, DeCicca et al. 2006) the search for other important determinants of smoking remains the focus of much current research.

In particular, the importance of social influences on adolescent choices (including health and educational choices) has been the subject of a large and growing literature in economics. Regarding health decisions, there is evidence that peers influence individual decisions to use drugs (Gaviria and Raphael 2001, Kawaguchi 2004), drink alcohol (Kremer and Levy 2003, Kawaguchi 2004, Lundborg 2006), and initiate sex (Fletcher 2007). Several authors have also found evidence of peer influences in adolescent smoking decisions (Powell et al. 2005, Gaviria and Raphael 2001, Clark and Loheac 2007, Lundborg 2006).

Unfortunately, credibly estimating peer influences on individual health decisions is fraught with difficulty, and most previous work has been unable to address critical challenges in identifying a primary parameter of interest—endogenous social effects, where an individual's propensity to behave in some way varies with the behavior of the group (Manski 1993). Determining this effect is policy relevant because it suggests that (1) the composition of peer groups are important determinants of adolescent smoking decisions and (2) interventions that decrease the smoking propensities of individuals will decrease the smoking propensities of their peers.

There are several large challenges in addressing whether endogenous effects exist for a particular behavior. First, peer groups (assumed to be classmates in this paper) are endogenously determined by parental choices (e.g. residential choices, private school enrollment). Second, there are likely to be group-level unobservables that influence individual and classmate choices simultaneously. Third, linear empirical models of social interactions suffer from identification problems for the coefficient that reflects endogenous effects (Manski 1993, Brock and Durlauf 2001), and instrumental variables or other assumptions are needed for identification.

In order to address these issues, I use the Add Health dataset and an instrumental variables/fixed effects methodology that compares students in different grades within the same high school who face a different set of classmates and classmates' decisions. I use school-level fixed effects to eliminate the majority of group unobservables as well as reduce the problem of the endogeneity of school (Hoxby 2000, Hanushek et al. 2003). I then use several alternative sets of instrumental variables that have been suggested in previous research in order to identify the endogenous effect. Preferred specifications suggest that increasing the proportion of classmates who smoke by 10% will increase the likelihood an individual smokes by approximately 3 percentage points. Falsification tests suggest that the estimates are robust. Finally, I provide evidence that research that is unable to use the combined IV/FE approach likely estimates biased results of the effect of smoking decisions for adolescents.