Education to improve labor market opportunities: a systematic review
Systematic review
|Published
The Norwegian Labor and Welfare Administration (NAV) commissioned the Norwegian Institute of Public Health to summarize research on the effects of educational interventions aimed at improving labor market opportunities for adults, who either are out of the labor force or have low skills, on employment outcomes.
Key message
The Norwegian Labor and Welfare Administration (NAV) commissioned the Norwegian Institute of Public Health to summarize research on the effects of educational interventions aimed at improving labor market opportunities for adults, who either are out of the labor force or have low skills, on employment outcomes.
We searched electronic databases and grey literature sources to identify randomized-controlled trials (RCTs) and registry-based studies (RBs). International methods standards were followed to synthesize the evidence. We included seven RCTs involving about 74,000 participants. All but one Dutch study were conducted in USA, mostly during the 1990s. Most of the RCTs did not randomize participants adequately. Because of incomplete information in the studies, it was challenging to assess their methodological quality. The RCTs evaluated multi-component programs, which included education but also various work-related services. Here, we refer to the interventions as ‘educational programs’. The most common comparison was no program participation. We also present information on six Nordic RBs (with 383,566 participants), although drawing causal inferences from RBs is more difficult due to the RBs’ retrospective design and selection bias. Briefly, the results are:
- Educational programs may result in little to no difference on people’s employment probability up to five years follow-up.
- Educational programs may result in little to no difference on people’s total earnings, average total earnings, working hours, and job duration (weeks worked).
- It is uncertain whether educational programs, compared to no intervention, improve people’s employment rates, average monthly earnings, and average annual earnings.
- It is uncertain whether educational programs, compared with programs focusing on on-the-job training and job search assistance, improve people’s employment probability and average annual earnings.
Six Nordic registry-based studies suggest that education may improve employment probabilities for immigrants and sick-listed adults.
Summary
Background
The Norwegian Labor and Welfare Administration (NAV) commissioned the Norwegian Institute of Public Health to summarize research on the effects of educational interventions aimed at improving labor market opportunities for adults, who either are out of the labor force or have low skills, on employment outcomes. This systematic review aimed to summarize the effects of educational interventions aimed at improving labor market opportunities for adults, who are either out of the labor force or have low skills, on employment outcomes.
Method
A systematic literature search in major literature databases and grey literature sources was conducted in October 2018. We searched for randomized-controlled trials (RCTs) and registry-based studies (RBs) that evaluated the effects of educational programs on labor market outcomes. Two independent reviewers selected and critically appraised the RCTs and RBs. The certainty of the evidence from the RCTs was evaluated following the GRADE approach (Grading of Recommendations Assessment, Development and Evaluation).
Results
We included seven RCTs (12 publications with about 74,000 participants). We also identified 26 RBs and described findings from six Nordic RBs (383,566 participants). Four RCTs included unemployed adults, and three included both low-skilled and low-wage workers. The RCTs evaluated multi-component programs, which offered a mixed package of services consisting of educational services (classroom education/training) alongside various work-related services. Here, we refer to the interventions as ‘educational programs’. The outcome measures reported in the RCTs were: employment, earnings, working hours, job duration (weeks worked). We synthesized the results and conducted meta‐analyses when possible. Poor data reporting of number of participants and precision measures precluded more detailed analyses.
Employment
Comparison 1: Educational programs versus no intervention
Employment rate: At 1-year follow-up, assuming that 10 of 100 people are employed under no intervention, we observed no difference between educational programs and no intervention (3 RCTs). The analyses showed that three more people (between 1-4 more people) would be employed if they participate in educational programs at 2-years follow-up. A larger effect of four more people being employed (between 1-6 more people) was observed at 3-years follow-up. This difference decreased to one more person being employed (between 0-1 more person) at 4-years follow-up. No difference was observed at 5-years follow-up (very low certainty in the evidence).
Employment probability: At 1-year follow-up, assuming 933 per 1000 people are employed under no intervention, nine more people would gain employment (19 more unemployed to 37 more employed) if they participate in educational programs (1 RCT). At 2-years follow-up, the difference in favor of the educational program increased up to 36 more people gaining employment (between 12-66 more employed) (2 RCTs; very low certainty). At 3-years follow-up, assuming 797 per 1000 people are employed under no intervention, eight more people would gain employment (between 40 more unemployed to 56 more employed) if they participate in educational programs (1 RCT; low certainty). No difference was observed at 4-years follow-up (low certainty).
Comparison 2: Educational programs versus programs focusing on on-the-job training/job search assistance (OJT/JSA)
Employment probability: The analysis used data from four sites in one RCT. Assuming 822 per 1000 people are employed under OJT/JSA programs, the analyses showed that 25 fewer people would be employed if they participate in educational programs up to 5-years follow-up (very low certainty).
Overview of findings from the Nordic registry-based studies (n=6): In general, the Nordic RBs reported that educational programs resulted in positive employment effects. Four Norwegian studies reported favorable employment effects for educational programs. A Norwegian study showed that ordinary education did not lead to any employment effect for young people with reduced work ability in the long term. One Danish study concluded that ordinary education and wage-subsidized job training resulted in positive employment effects for sick-listed employees, whereas the two other programs, non-formal education and subsidized internships, yielded negative employment effects.
Earnings, working hours and job duration
We assumed earnings of 10,000 NOK/month as a baseline effect under no intervention to interpret differences between educational programs and no intervention. This assumption applies for all earnings measures.
Comparison 1: Educational programs versus no intervention
Total earnings: People with low skills would earn 5,000 NOK more (between 1,000 NOK to 10,000 NOK more) by taking part in educational programs at 1-year follow-up. Similar benefits were observed up to 5-years follow-up (low certainty). People with high skills would earn 400 NOK more (between 1,000 NOK less to 10,000 NOK more) by taking part in educational programs at 1-year follow-up (very low certainty); larger benefits of 1,000 NOK more in total earnings were observed at 2-5-years follow-up (low certainty).
Average total earnings: One RCT provided data for employed adults receiving welfare benefits. Participants in the educational program (community college) earned on average about 21,000 NOK (about $2,300 USD) less than peers who did not participate over 2.5-years follow-up (low certainty).
Average monthly earnings: One Dutch RCT reported that the average monthly earnings of low-skilled workers who participated in classroom training were 10.9% higher than those who did not participate in any training at 2-years follow-up (low certainty). One US RCT found no difference between educational programs and no intervention on single mothers’ average monthly earnings at 2-3-years follow-up (very low certainty).
Average annual earnings: People with low skills would earn 100 NOK/month less by participating in educational programs at 1-3-years follow-up. A positive difference of 300 NOK/month more was found at 4-6-years follow-up. Similarly, a larger difference of 500 NOK/month more was found at 7-9-years follow-up (very low certainty).
Comparison 2: Educational programs versus programs focusing on on-the-job training/job search assistance (OJT/JSA)
Average annual earnings: People would earn 200 NOK/month more if they participated in OJT/JSA programs compared to educational programs at 10-15-years follow-up (very low certainty).
Working hours: Participants in the educational programs worked on average 13.3 hours more per week (between 3-24 hrs more) than peers under no intervention at 1-year follow-up (low certainty). This effect increased up to 18.3 hours more (between 6-31 hrs more) at 2-years follow-up (low certainty).
Job duration (weeks worked): Participants in educational programs worked 0.39 weeks more than those in no intervention at the 1-year follow-up. This effect was 0.35 weeks more worked at 2-years follow-up (low certainty for both estimates).
Conclusion
Overall, the evidence suggests that educational programs may result in little to no difference for most of the labor market opportunities for unemployed adults or low-wage workers. With respect to employment rates, the results show that it is uncertain whether participants in educational programs have better employment rates compared to non-participants. We note that these findings are based on RCTs, mostly carried out in USA during the 1990s, that have serious methodological weaknesses. Additionally, their transferability to the Norwegian context is limited. The six Nordic RBs, with data from the 2000s, similarly indicated small positive employment effects, and might be used in evidence-informed policymaking, taking into consideration the limitations of non-experimental studies.
Taken together, the results from the RCTs and the RBs suggest small positive effects of educational programs, but there is a need for additional research.
