We study the US labor market transitions using a latent variable approach, explicitly modeling the persistent misclassification process and the non-Markovian nature of the underlying true labor force dynamics. A closed-form global identification for misclassification probabilities and labor transition probabilities is established through an eigenvalue-eigenvector decomposition. Contrary to existing studies, our empirical results suggest that the observed data have understated the true mobility in labor force statuses after we account for persistence in both the misclassification errors and the latent true labor force dynamics.

Do peers in childhood influence the development of noncognitive skills? Despite a large literature on peer effects, this question remains largely unanswered. In this paper, we use Big Five Inventory measures of personality traits linked to the administrative records of children randomly assigned to primary school classrooms. We find that children exposed to “left-behind” classmates whose parents have temporarily migrated away for work become less conscientious, agreeable, and emotionally stable. These impacts mirror the traits of their left-behind peers with whom children become friends.