This paper examines the inuence of selection after the gate on student graduation rates. We use the variation in academic dismissal (AD) policies as an instrument for the endogenous student dropout level. Using a parametric 2SLS estimation and a nonparametric IV regression, the instrumented variation in student dropout is used to carve out the impact on graduation rates. The IV outcomes on Dutch program level data suggest that: (i) student dropout increases with higher academic dismissal thresholds, and (ii) programs which are more selective in the rst year have higher student graduation rates. The results also indicate that student satisfaction, the percentage of female students in the study program and achieved level of the study program are associated with higher graduation rates.
JEL-Classi cation: C36; I21
Keywords: Higher education; Academic dismissal policy; Student dropout; Completion rates; Instrumental Variable estimation.