This paper derives school-to-work transition pathways in the United States and Europe between the late 1990s and the early 2000s. To do so, it uses Optimal Matching, a technique developed to sequence DNA. The key advantage of using this technique is that, rather than focusing on a specific point in time or a single destination, such as employment, inactivity or unemployment, they convey information on all activities undertaken by youth over the transition period, their sequence and their persistence. Strong similarities are found between the United States and Europe. However, pathways in the United States are characterised by significantly more dynamism than in Europe: youth in employment tend to change jobs more frequently while inactive or unemployed youth are more likely to experience several short spells rather than a single long one. School-to-work transition pathways in the United States also involve less time spent in unemployment than in Europe. The share of school-leavers involved in pathways dominated by employment is larger in the United States than in Europe and non-employment traps are less frequent in the United States. The most successful European countries in terms of school-to-work transitions are those where apprenticeships are widespread. On the other hand, European countries with a high incidence of temporary work among youth have a significantly smaller share of youth belonging to pathways dominated by employment and a larger share of youth in pathways characterised by frequent job changes separated by long unemployment spells. At the individual level, qualifications, gender, ethnicity and motherhood are found to influence the probability of disconnecting from the labour market and education for a prolonged period of time. Overall, the analysis shows the potential of Optimal Matching as a descriptive tool for the study of school-to-work transitions. It also tentatively explores how pathways obtained through Optimal Matching could be used for further analysis to draw policy-relevant conclusions. At present, data availability appears to be the main barrier to fully exploiting this novel technique.
