This paper uses panel data and econometric methods to estimate the incidence and the dynamic properties of overskilling among employed individuals. The paper begins by asking whether there is extensive overskilling in the labour market, and whether overskilling differs by education pathway. The answer to both questions is yes. The paper continues by asking whether overskilling is a self-perpetuating labour market state (state dependence), and whether state dependence differs by education pathway. The paper uses a dynamic random effects probit which includes Mundlak corrections and it models the initial conditions following Heckman's method. It finds that there is extensive overskilling state dependence in the workplace, and to the degree that overskilling can be interpreted as skills underutilisation and worker-job mismatch, this is an important finding. Overskilled workers with a higher degree show the highest state dependence, while workers with vocational education show none. Workers with no post-school qualifications are somewhere between these two groups. The finding that higher degree graduates suffer the greatest overskilling state dependence, combined with the well-established finding that they also suffer the highest overskilling wage penalty, offers an alternative and useful perspective to compare the attributes of vocational and degree qualifications.