Causal effects of mathematics

This paper exploits that students at age 16 in Norway are randomly selected into one compulsory exit exam in either mathematics or languages. A few days before the actual exam day, the students are notified about exam subject. The students have an intensive preparation period, and examination in mathematics relative to languages is found to decrease dropout from high school, increase enrollment in higher education, and increase enrollment in natural science and technology education programs. Overall, the causal effects seem to be somewhat stronger for males than for females, but the analysis indicates that gender differences interact in complicated ways with prior skills in mathematics. We explore several mechanisms that might contribute to the findings.