To obtain data on these early labor market periods, individuals with non-interviews in the relevant periods, individuals missing the enrollment data required to date their labor market entry, and individuals who did not have a first labor market entry (as defined above) in the 1979-1982 period had to be dropped; obviously, these restrictions generate large reductions in the available sample, because they tend to exclude both the oldest and youngest members of the NLSY cohort. GN present evidence suggesting that these sample selection rules hinge largely on the age of respondents-although obviously schooling decisions also play a role-and therefore may not generate substantial biases.

Some observations are also lost because of missing or inconsistent data on early labor market experiences, or because respondents were not observed in a job during the survey week at any time during the five-year post-schooling period (which eliminates only about six percent of the sample).

Other than this latter restriction, imposed so that some job information is available, there is no lower limit on the amount of time they had to have worked during that period. Finally, because the outcome of interest is adult wages, attention is restricted to those working for a wage on the CPS job in the NLSY in 1992, which refers to the main job held in the previous week. payday loans