The Impact of Grade Retention on Juvenile Crime

Abstract

Using detailed administrative and individual data on schooling and crime records from Chile, we estimate the effect of grade retention between 4th and 8th grade on juvenile crime. We base our research on the rule which specifies that students who fail more than one subject must repeat the year. We present two empirical strategies to address the strong evidence that the forcing variable is – locally – manipulated. First, we follow Barreca, Guldi, Lindo, and Waddell (2011) in implementing a donut-hole fuzzy regression discontinuity design (FRD). Second, we extend the approach developed by Keele, Titiunik, and Zubizarreta (2015) to implement a method that combines matching with FRD. These two methodologies deliver similar results and neither show a statistically significant effect on a placebo test. According to our results, grade retention increases the probability of juvenile crime by 1.6 percentage point (pp), an increase of 33% (higher for males and low SES students). We also find that grade retention increases the probability of dropping out by 1.5pp. Regarding mechanisms, our findings suggest that the effect of grade retention on crime does not only manifest itself indirectly as a result of its effect on dropping out. Furthermore, the effect of grade retention on crime is worsened when students switch schools right after failing the grade.