This paper assesses the dynamics of treatment effects arising from variation in the durationof training. We use German administrative data that have the extraordinary feature that theamount of treatment varies continuously from 10 days to 395 days (i.e. 13 months). Thisfeature allows us to...