I. INTRODUCTION The problems of teenage alcohol use are well documented in the public health and health economics literatures. Adolescent alcohol consumption is associated with adverse health consequences, crime, traffic fatalities, out-of-wedlock teenage childbearing, lower socioeconomic status, and poorer quality marriages (Cook and Moore, 1993; Dee, 1999, 2001; Grossman et al., 1993; Kenkel and Ribar, 1994; National Highway Traffic Safety Administration, 2000; Rees, Argys, and Averett, 2001; Sen, 2003). Underage drinking is also not uncommon. According to the 2003 Youth Risk Behavior Study, over half of high school students report consuming at least one drink in the past 30 d (Youth Risk Behavior Study, 2003). Because alcohol consumption patterns often emerge during individuals' early teenage years, understanding the determinants and consequences of adolescent drinking behavior is important (DeSimone and Wolaver, 2006; Koch and McGeary, 2005). Recent empirical studies examining the relationship between teenage alcohol consumption and schooling have produced mixed results. Several studies have found that alcohol consumption reduces investments in schooling (Cook and Moore, 1993; Koch and McGeary, 2005; Mullahy and Sindelar, 1989; Williams, Powell, and Wechsler, 2003; Wolaver, 2002; Yamada et al., 1996), but other recent work (Dee and Evans, 2003; Koch and Ribar, 2001; Duarte and Escario, 2005) casts doubt on these findings, suggesting that alcohol consumption has only a modest effect or no effect on schooling. Each of these studies has stressed the importance of carefully addressing unmeasured heterogeneity, which may mitigate estimated human capital effects. Despite a growing body of literature on the effects of teenage drinking on future educational attainment, less attention has been paid to the effects of adolescent alcohol consumption on contemporaneous academic achievement. While examining the effect of teenage drinking on school attainment is important and informative, focusing on this outcome alone may understate the full effects of drinking. If alcohol consumption negatively affects precollege academic performance by diminishing cognitive abilities or by reducing the efficiency of producing effective study time, these adverse academic spillovers may have important direct effects on future human capital accumulation and earnings. Betts and Morrell (1999, p. 269) note that grade point average (GPA) reflects human capital acquisition at a time when young adults are close to permanent entry into the labor force. Moreover, there is evidence to suggest that high school GPA is an important determinant of earnings (Grogger and Eide, 1995; Rose and Betts, 2004) and academic performance in college (Barron and Norman, 1992; Cohn et al., 2004). This study uses a national data set not yet exploited in the alcohol-human capital literature, the National Longitudinal Study of Adolescent Health (Add Health), to examine the relationship between teenage binge drinking and three measures of academic performance: pre-college GPA, receipt of an out-of-school suspension, and having an unexcused absence from school. Ordinary least square (OLS) estimates consistently show that there is a negative relationship between adolescent binge drinking and academic performance. However, after exploiting multiple waves of the Add Health data and including individual fixed effects, the estimated effect of binge drinking on academic performance falls substantially. These findings suggest that previous estimates of the effect of binge drinking on schooling may be overstated. II. THEORETICAL FRAMEWORK An adolescent is assumed to maximize utility, U(A, S, L)--which is a function of alcohol consumption (A), academic performance (S), and leisure (L)--subject to time, budget, and human capital production constraints. (1) The schooling production function is given by: (1) S = [Florin] (m, pe, a, q, h, t, z), where m is student motivation, pe is parental effort and involvement, a is student ability, q is school quality, h is health, t is time spent studying, and z are individual- and community-level tastes. …
Read full abstract