Association for Public Policy Analysis & Management

Sleep is an important component of cognitive performance, but current school start time policies require students to wake up far earlier than health experts recommend. In this paper, we analyze the effect of school start times on academic performance. Sleep patterns are determined in part by sunrise times, which vary over time zones. Since school start times do not fully reflect this, we instrument for the amount of sunlight before school with the time zone boundary in Florida and Tennessee. Using individual-level data, we find that moving relative start times one hour later increases math scores by 0.04 standard deviations for younger children, but by 0.10 SDs for adolescents. The age profile of the start time effect corresponds to differential entry into puberty by gender, as defined by the Tanner Scale. We also find that later start times increase reading scores by about 0.07 SDs for children of all ages. In both reading and math, later start times have a similar impact on test scores across the distribution of socioeconomic status and baseline ability. Later start times are associated with a 24% reduction in suspension rates for adolescents, which we take as tentative evidence that executive control may be an important causal avenue through which later start times improve academic outcomes for adolescents. We conclude that school districts may be able to significantly boost academic achievement by adjusting the order in which schools for different aged children open.