This study examines long-term effects of a minimum wage increase using an innovative identification strategy based on categorising workers according to their predicted marginal revenue products. It finds that the increase had a large and persistent disemployment effects on low-paid workers and that it triggered substitution toward more productive workers. As a consequence, the sub-minimum workers as a group lost average earnings, hours and employment compared to other workers. The adverse employment effect occurred both through a higher probability of transition from employment to non-employment and through a decreased probability of transition from non-employment to employment.