Pré-publication, Document de travail: Individual risk attitudes are widely used in order to predict decisions regarding education. These uses of risk attitudes as a control variable for education decisions, however, have been criticized due to potential reverse causality. The causality between risk aversion and education is not clear, and it is hard to disentangle the different directions. We here investigate the causal effect of education on risk aversion by looking at the 1973 British Education Reform. With the educational reform in 1973 when the end of compulsory education increased from 15 to 16, it is possible to see the average treatment effect of education reform on the individual risk attitude for different cohort groups. We find that years of schooling increase risk aversion level via IV2SLS, which is contrary to the existing literature to our knowledge. This result is especially stronger for those with lower education. We suggest that in early education, education makes individuals more risk averse, whereas in more adult education such as tertiary education, years of schooling diminish risk aversion as suggested in other literatures. In addition, this negative causal effect of education on risk aversion could relieve our concerns about the endogeneity/reverse causality issue when using risk aversion as an explanatory variable for education decisions, because the sign would still credible as coefficients are underestimated.