| The three chapters of this dissertation are devoted to analyze how young adults make decisions, and the consequences of these decisions for their adult lives. In Chapter 1, I investigate the causal effects of consuming illegal drugs on educational attainment, employment and wages. To identify these effects, I develop and estimate a structural dynamic model with simultaneous decisions on drug consumption, school attendance, and participation in the labor market, and allow these decisions to affect future opportunities. My analysis focuses on male respondents in the National Longitudinal Survey of Youth, 1997. I find that the main effect of drugs is through wages. Contrary to findings in the literature, non-drug users have higher wages than marijuana and/or hard drug users. This effect is small for individuals who consume marijuana in very low doses, but it is large for frequent drug users. I also find that school attendance and employment are negatively affected by drug consumption, though the effects are milder. In Chapter 2, I examine the relationship between investments in human capital and long-term career paths. Specifically, I allow individuals to choose between two types of schooling, which I term basic schooling and general schooling. These two types of schooling require different investments in human capital, and give rise to vastly different career choices. In addition to being in school, individuals can be engaged in work activity. I use data from the National Longitudinal Survey of Youth, 1979. The results shed new light on the career choices of young men. I find that the two channels of education lead individuals to very different career paths, and that the return to these two types of schooling are different across occupations. I also find that occupation-specific experience is very important; the return to experience is highest in the occupation in which experience was acquired, and dramatically lower in other occupations. In Chapter 3, I examine evidence of peer effects in risky teenage behaviors such as smoking, drinking and taking drugs. I use two datasets, namely, the National Longitudinal Study of Adolescent Health (Add Health), and the Los Angeles Family and Neighborhood Survey (L.A.FANS). My approach is to use a rich set of covariates and neighborhood fixed effects to control for a wide variety of individual, friend, and neighborhood characteristics. First, using Add Health data, I estimate the probability of friendships among teens based on observable characteristics of individuals, their friends, and their environment. I use these estimated friendship probabilities to pre-diet friendship in the L.A.FANS dataset. Second, using predicted friendship probabilities to weight the influence of different friends, I estimate the effect of peer behavior on an individual's behavior in the L.A.FANS data. Finally, because individuals may select friends based partially on their participation in risky behaviors, I substitute the actual behavior of friends with the predicted probability of this behavior using Add Health data. I find that, even with a rich set of controls, peers have a small but significant effect on individual behavior. However, when I use predicted behavior, the effects become very small and, in most cases, insignificant. |
