Abstract
This paper investigates the effect of peer smoking on individual smoking among youths in 10 countries that participated in the European Global Youth Tobacco Survey (GYTS). I control for endogeneity in school selection and unobserved school-level characteristics through the use of school fixed-effects. I use instrumental variables to address the simultaneity in peer and individual behaviours. Identification arises by comparing students in different classes within the same school. On average, an increase in the share of classmates who smoke by 10 percentage points increases the probability that an individual in that class will smoke by 3 to 6.9 percentage points. The results imply that any policy intervention such as anti-smoking messages, smoking bans, or higher cigarette prices will be even more cost-effective because of the social multiplier effect of peers – policies affecting some individuals in a group will generate spillovers to others through the peer effect.