Insights / Research BriefApr 07, 2023

What Works for the Unemployed? Evidence from Quasi-Random Caseworker Assignments

Anders Humlum, Jakob R. Munch, Mette Rasmussen
Two years after job loss, jobseekers assigned to classroom training programs tend to work 25% more, whereas on-the-job training programs appear to have no robust employment effects. These results, which are driven by post-program effects, challenge the existing literature.

Labor market policies, which include programs such as vocational training, job search aid, and wage subsidies, are of increasing importance for ensuring a productive workforce amid ongoing structural changes in the labor market. Despite this, there is limited evidence regarding their effectiveness. Workers who opt into training programs likely differ systematically from those who do not, casting doubt on the results of studies that simply compare the outcomes of workers who do and do not complete training. This study overcomes this limitation by using data on jobseekers who are quasi-randomly matched with caseworkers to assess the impacts of a particular labor market policy in Denmark — classroom training. 

The authors use administrative data from Denmark covering jobseekers who lost their jobs between 2012 and 2018. Importantly, for unemployed jobseekers to receive UI benefits from the Danish government, they must meet with a caseworker at a job center to receive assistance with their job search and assignment to a training program. Jobseekers are assigned caseworkers essentially randomly (based on their day of birth), and caseworkers differ in their tendencies to assign jobseekers to different types of training programs, with some caseworkers more likely to assign jobseekers to classroom training programs and others more likely to assign jobseekers to programs that provide training on the job. The authors exploit this in their research design and compare the employment outcomes of jobseekers from the same job center and year who, due to their day of birth, receive different counseling. They find the following: 

  • Jobseekers who are assigned to classroom training tend to work more as a result. These employment gains grow steadily over time, stabilizing at about 25 hours more per month two years after their initial job loss, equivalent to a 25% increase relative to before their job loss. 
  • By contrast, on-the-job-training programs, such as employment programs with wage subsidies, do not lead to employment gains.
  • These results diverge from earlier studies that rely on observable characteristics of jobseekers and often conclude that classroom training has deleterious effects on employment. By contrast, this work accounts for selection based on unobserved characteristics when evaluating labor market policies; for example, jobseekers who face worse employment prospects are more likely to opt into training.

In the next part of the paper, the authors aim to uncover the mechanisms driving the effects revealed in their analysis. They find the following:

  • The benefits of workforce training are driven primarily by the positive effects accrued by participants who complete the programs, rather than by jobseekers who simply exit unemployment upon commencing training. This suggests that classroom training increases employment by providing job seekers with skills that are valued in the labor market.
  • Assignment to classroom training especially increases employment outside jobseekers’ original occupations, providing further support for the conclusion that workforce training helps workers find jobs through the provision of new skills.
  • The employment effects of classroom training are driven by participants’ more successful job applications rather than by their intensified job searches, again underscoring the role of skill acquisition. 

The authors conclude by exploring how these effects vary across different types of workers in order to offer insights relevant for policy:

  • Jobseekers who are employed in occupations that are more exposed to offshoring have higher employment gains from classroom training.  By quarter seven after their initial job loss, high-risk jobseekers gain 55 hours of employment per month from assignment to classroom training. This gain corresponds to 50% of their pre-job-loss level of employment. 
  • In contrast, jobseekers at low risk of offshoring derive much lower employment gains from assignment to classroom training. By quarter seven after their initial job loss, the gains for low-risk job seekers are not statistically significantly different from zero. 
  • Taken together, these results suggest that a cost-effective way to close the employment gap is to redistribute classroom training from low-risk to high-risk jobseekers. Note that this counterfactual scenario corresponds to assigning 25% of all job seekers to classroom training, compared to today’s 39%. Hence, this policy would lower total spending on classroom training programs while bolstering their effect on employment.

This paper provides novel evidence for the effectiveness of classroom training for helping displaced workers regain employment. From a methodological standpoint, this study illustrates the potential pitfalls in assigning causality in studies lacking at least quasi-randomness. For policymakers, this research offers a tool for closing the employment gap among job seekers affected by offshoring.