Krueger, A., & Rouse, C. (1994). New evidence on workplace education (No. w4831). Cambridge, MA: National Bureau of Economic Research.
- The study’s objective was to examine the impact of employer-based workplace education on wage growth at two mid-sized companies in New Jersey, one in the manufacturing sector and the other in the service sector.
- In this nonexperimental study, the authors used employee survey data and administrative records from the participating companies and the community college that provided the training to compare the wage growth of workers who participated in training with the wage growth of workers who did not participate.
- The study found that employees in the manufacturing company who participated in occupational training had higher wage growth than those who did not participate in the training.
- The quality of causal evidence presented in this report is moderate because it was based on a well-implemented nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the employer-based workplace education program, but other factors might also have contributed.
Employer-Based Workplace Education
Features of the Intervention
Two mid-sized companies in New Jersey, one in the manufacturing sector and the other in the service sector, implemented the training program. It was designed by adult education specialists at a local community college and funded by a federal grant. Training ran from October/November 1992 through February 1994. Courses were provided on site by teachers who specialized in adult education. Classes met twice per week for two hours and were taught in five 8- to 12-week sessions. Employees were paid their regular wages during class time, and signing up was voluntary. Classes were taught during regular shift times, so employees could participate only if their absence would not disrupt the work flow. Subjects covered in the training included basic skills, such as reading, writing, math, and English as a second language. In addition, the training at the manufacturing company included occupational skills classes, such as blueprint math and blueprint reading.
Features of the Study
The authors used regression analyses to compare the wage growth of workers who participated in training with the wage growth of workers who did not participate. The analyses were conducted separately by company and included controls for age, tenure, gender, race, education, and wages in 1991. The sample consisted of 503 workers from the manufacturing company and 220 workers from the service company who had started with their company before 1992 and remained with them through early 1994 or later. The study used data from company administrative records, employee surveys, and administrative records from the community college that organized and designed the training program.
- The study found that participants in occupational skills classes at the manufacturing company had significantly larger wage growth than nonparticipants at the manufacturing company.
Considerations for Interpreting the Findings
The study noted large pre-program dissimilarities between treatment and comparison groups on hourly wages, age, tenure, years of education, and percent female. Before the program, training participants, on average, earned less per hour, were younger, had less seniority, had higher levels of education, and were more likely to be female than nonparticipants. Even though the analysis included statistical controls for these measures, these controls might not be able to fully account for existing differences of this magnitude. It is possible that varying levels of pre-intervention wages (for example), rather than the intervention, could explain any observed differences in outcomes between the groups.
Causal Evidence Rating
The quality of causal evidence presented in this report is moderate because it was based on a well-implemented nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the worker training program, but other factors might also have contributed.