Do Performance Targets Affect Risky Behaviour? Evidence from Discontinuities in Test Scores in England∗

Performance targets are ubiquitous in all areas of an individual’s life, such as education, jobs, sport competitions and charity donations. In this paper I study the effect of meeting performance targets in school tests on the probability that students subsequently engage in risky behaviour. This is helpful to assess whether behavioural channels such as motivation and effort by students, parents and teachers, that meeting a target may lead to, play a role in influencing risky behaviour. I address potentially spurious correlations between test scores and students’ behaviour by exploiting a regression discontinuity design in test scores and a linked dataset with information on students in compulsory education in England. I find that meeting targets that the government set for students at age 11 has a negative but not significant effect on proxies for risky behaviour, such as the probability of unauthorised absence from school or of a police warning to students’ parents. In addition, the effect varies by students’ ability, gender, parents’ education level and type of risky behaviour. In contrast, linear probability model estimates of the effect are negative and significant. The empirical evidence informs policy decisions about education and support to young people by suggesting i) no major behavioural implications of performance targets in tests and ii) that naive estimates not correcting for unobservables are spurious and may lead to inaccurate policy. JEL Classification: C21, I20, I21, I28