Do early job losses permanently reduce the earnings and career prospects of young
workers? Simple estimates may overstate the true effects of early displacements, especially if less able
workers sort into firms with high turnover rates. The bias from initial assignment of workers between
firms is compounded by biases from selection within firms, which arise if employers selectively
displace their least able workers, or if workers move voluntarily to take better jobs. This paper uses
longitudinal social security data on German apprentices and their training firms to obtain estimates
of the long-term effects of an early job loss that account for nonrandom assignment between firms
and selection within firms. I use differences over time in the fraction of graduating apprentices that
are retained by the training firm as an instrument for job displacement. These should reflect
exogenous changes in firm-specific labor demand that are independent of individual ability or
permanent firm characteristics. Using this strategy, I find that wage losses from leaving the training
firm at graduation are initially strong but fade within the first five years in the labor market. The
results also confirm an important influence of voluntary mobility and of initial sorting matching
trainees to firms. Both of these factors are likely to confound results of previous studies of early job
mobility lacking information on the demand side. |
