We study the impact of early cannabis use on the school to work transition of young men. Our empirical approach accounts for common unobserved confounders that jointly affect se- lection into cannabis use and the transition from school to work using a multivariate mixed proportional hazard framework in which unobserved heterogeneities are drawn from a discrete mixing distribution. Extended models account for school leavers' option of returning to school rather than starting work as a competing risk. We find that early cannabis use leads young men to accept job offers more quickly and at a lower wage rate compared to otherwise similar males who did not use cannabis. These effects are present only for those who use cannabis for longer than a year before leaving school. Overall, our findings are consistent with a mechanism whereby early non-experimental cannabis use leads to greater impatience in initial labor market decision-making.
multivariate duration models discrete factors cannabis use job search wages