Charlotte Cabane

We analyse the effects of playing music, or doing sports on education and health outcomes of adolescents. After identifying adolescents who play music, do sports, or both, in the German Socio-Economic Panel, we use matching procedures to estimate causal effects. We find that playing music instead of doing sports fosters educational outcomes by about 0.1 standard deviations. Effects are stronger for girls, and for children from more highly educated families. Doing sports improves perceived health more strongly than playing music. Engaging in both activities, music and sports, improves educational outcomes by about 0.2 standard deviations and reduces smoking by about 10 percentage points compared to engaging in just one activity. Adolescents who engage in music spend less time watching TV or playing computer games, but more reading books. The robustness of the results is examined with respect to the identifying assumptions, including non-affected outcomes, a formal sensitivity analysis, and instrumental variable estimation. These checks do not reveal any serious problems.