Tracking Adolescent Maturation using Risky Behaviors

讲座内容： Adolescent riskybehavior is an enduring source of policy concern. Adolescents engage in risky behaviors,according to one plausible explanation, because of a lack of self-control; theygain self-control as they mature, causing them to quit or moderate theiractions. While this explanation is difficult to test given the maturationprocess is not observed directly, behavioral changes by age can be used to shedlight on its underlying nature. To that end, this paper constructs the firstbehavioral framework to analyze both the effect of maturation on riskybehaviors and the timing of maturation. In measuring the effect of maturation,situational changes among observationally similar individuals could inprinciple be differenced out as in a difference-in-differences analysis if thetiming of the treatment – when the adolescents mature – were known. Inpractice, it is not. To solve this identification problem, I develop a two-partstrategy that allows me to recover both the effect of maturation and its timingfor each individual. First, assuming the timing of maturation is known, I applya standard difference-in-differences analysis, using relative changes betweenthe now-mature individuals and their peers to identify the maturation effect;second, given the maturation effect, the timing is identified as a structuralbreak for each individual. In this way, the two parts complement each other toachieve joint identification. Based on data from the National LongitudinalSurvey of Youth 1997, I find that the estimated maturation age distributionsfor males and females both peak at age 21, and are right-skewed. As one policy-relevantapplication of the framework and estimates, I decompose the age profiles as a guidein setting the legal drinking age. Here I show that a counterfactual increasein the legal age from 21 to 23 could, assuming perfect enforcement, lead to a6.8% decrease in the binge-drinking rate for females, in contrast to the actualincrease of 1.5% (on a base of 41%). My analysis therefore suggests that eventhough the current legal drinking age in the United States of age 21 is on theconservative side, a further increase may be justified by the late maturationof many young people.