Association for Public Policy Analysis & Management

Over the past few decades, teenage risky behaviors, such as smoking, substance abuse, unprotected sex and violence have been a rising concern in the United States. This is because of the consequences associated with them, which are both short run safety or learning (Dee and Evans, 2001; Card and Lemieux, 2001) and long-run income and longevity (Farrell and Fuchs, 1982; Bhattacharya and Currie, 2001). In a 2016 survey by CDC, teen pregnancy has become one of the major reasons for high school dropout for girls. Only 50% of the young mothers receive a high school diploma by 22 years of age, thus making them more likely to be unemployed (Centers for Disease Control and Prevention, 2017). Despite substantial importance of understanding teenagers’ risky behaviors and its implications, surprisingly little research has been done in this area.

This paper aims to fill part of this gap in literature by examining the relationship between school characteristics (school size in this case) and risky teen behavior. Schools play a pivotal role in regulating teenagers’ behaviors. Research suggests attendance at schools reduces teenagers’ risky behaviors. (Jacob and Lefgren, 2003; Black, Devereux and Salvanes, 2008). A small school may generate impacts on student behaviors through various channels, such as increased graduation rates (Schwartz, Stiefel and Wiswall, 2013, 2015; Bloom, Thompson and Unterman, 2010), better student achievement (Howley, Strange and Bickel, 2000; McMillen, 2004) and long-term success in the educational system and the labor market (Humlum and Smith, 2015). Attending a small school may decrease the probability of engaging in risky behaviors through increased educational attainment such as achievement and graduation rates.

Using student-level administrative and survey data from New York City public schools, and Medicaid claims data, the paper evaluates the impact of small high schools on risky student behaviors and related health problems: teenage pregnancy, violence injury and mental and substance use disorders. It uses distance between student residence and school as an instrument for endogenous school enrollment, using a two-sample- instrumental-variable approach.

Preliminary results suggest that girls (between ages 15-18) enrolled in old small high schools (opened before 2003) in contrast to new small schools (opened in 2003 or after) or large schools are less likely to get pregnant or diagnosed with mental disorders. Similarly, boys in old small high schools are less likely to be diagnosed with mental and substance use disorders.

These findings yield crucial insights for state policymakers, school districts, and local education and health agencies with regards to the current small school policies. A closer look at school characteristics imply smaller schools size, higher per student expenses and lower student-teacher ratio as contributions to the success of old small high schools. Lower teacher quality also potentially explains the disparity between new and old small schools.