Citation
Andersson, F., Holzer, H. J., Lane, J. I., Rosenblum, D., & Smith, J. (2013). Does federally-funded job training work? Nonexperimental estimates of WIA training impacts using longitudinal data on workers and firms (Discussion paper no. 7621). Bonn, Germany: IZA. [Dislocated Workers ONLY]
Highlights
- The study’s objective was to examine the impact of the Workforce Investment Act (WIA) dislocated worker program’s training services on the employment and earnings of dislocated workers.
- The authors used a regression model with inverse propensity weights, to compare the earnings and employment outcomes of WIA-registered dislocated workers who received training services to the outcomes of those who were registered in WIA but did not receive training services.
- The study found that the treatment group earned significantly less than the comparison group in the three years after WIA registration. Individuals in the treatment group earned an average of $5,567 and $5,227 less than those in the comparison group in State A and State B, respectively.
- The quality of causal evidence presented in this report is moderate because it was based on a well-implemented nonexperimental design; this is the highest causal evidence rating possible for a nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the training services received, but other factors might also have contributed
Intervention Examined
Workforce Investment Act (WIA) Program Training Services
Features of the Intervention
The WIA dislocated worker program training services involved classes in occupational skills and on-the-job training provided by private firms. Selection into the training services could have occurred in a variety of ways, including referral from a training provider or program staff member, individual choice, or natural progression through prior services offered by WIA. There were no specific eligibility criteria for registering in the WIA program itself.
Features of the Study
The study implemented a nonexperimental analysis of the employment and earnings of WIA dislocated worker program participants in two states. State A, a medium sized state located on the East Coast, had 10,836 WIA dislocated worker program participants of whom 4,347 received the training services. State B, a large state located in the Midwest, had 28,246 WIA dislocated worker participants of whom 16,187 received training services. Using data from the Workforce Investment Act Standard Record Data (WIASRD) and Longitudinal Employer-Household Dynamics (LEHD), the authors estimated a regression model using inverse propensity weighting that compared the treatment group of participants who received training services to the comparison group of participants who did not receive training services. The comparison group was selected to match the treatment group in terms of demographic characteristics and prior employment and earnings.
Findings
- In the three years after WIA registration, treatment group members in State A and State B earned an average of $5,567 and $5,227 less, respectively, than those in the comparison group.
- In State A, the treatment group earned significantly less in quarters 1 through 6 and quarter 9 than the comparison group. • In State A, the treatment group was significantly less likely than the comparison group to be employed in quarters 1 to 3, but was significantly more likely to be employed in quarters 6 to 12.
- In State B, the treatment group earned less in quarters 1 through 7, compared to the comparison group. However, in quarters 10 through 12 the treatment group earned more, on average, than those in the comparison group.
- In State B, the treatment group was less likely to be employed than the comparison group in quarters 1 through 7, but was more likely to be employed in quarters 11 and 12.
Considerations for Interpreting the Findings
The program was designed such that WIA participants would receive core and intensive services, followed by training if they were still not employed. However, in practice, some participants received training without receiving core or intensive services. Thus, the treatment in this study can be understood as training plus the opportunity to receive core and intensive services, whereas the comparison condition is the opportunity to receive core and intensive services only.
The authors made a large number of comparisons in the earnings and employment domains without adjusting for these multiple comparisons. A large number of comparisons increases the probability that some impacts are found to be statistically significant by chance. After CLEAR adjustments, the impact of training services on third-year earnings in State A was no longer statistically significant.
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 is the highest causal evidence rating possible for a nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the training services received, but other factors might also have contributed.