S9 Ep36: Helping the over-50s find work

Episode Summary

Executive Summary: This episode examines a Dutch randomized trial of STEP, a job-search assistance program for unemployed workers over 50. The study finds the intensive program improved job-finding, raised job quality modestly, and paid for itself for government, though benefits were strongest for more employable participants and were likely recession-specific rather than a fix for structural labor-market change.

Main Topics: Older workers face a tougher re-entry into employment (Priority: 5/5): The discussion opens with the labor-market disadvantage faced by over-50 workers, who are more likely to experience long unemployment spells and lower job-finding rates than younger job seekers. Why older job seekers struggle (Priority: 5/5): Boss van der Klaauw explains the mix of factors: generous unemployment insurance, employer preferences for younger workers, and the difficulty of starting over late in a career. The STEP program and its design (Priority: 5/5): The Dutch government-funded STEP intervention was a high-intensity job-search assistance program that combined group sessions, individual coaching, and social-network activation for older unemployed workers. Experimental evaluation and participation (Priority: 4/5): The policy was evaluated through a randomized treatment-control experiment among people over 50 applying for unemployment insurance in the Netherlands, with substantial take-up among those invited. Effects on employment, earnings, and program cost (Priority: 5/5): The program increased job-finding, slightly improved earnings through faster re-employment, and reduced unemployment-insurance spending enough to make it cost-effective for government. Who benefited most and job quality effects (Priority: 4/5): The strongest gains were among more educated, more employable participants; the program also improved job quality by increasing permanent contracts and working hours. Limits of applicability in an AI/structural change era (Priority: 4/5): Van der Klaauw argues STEP addresses frictional unemployment in recessions, whereas AI-driven displacement may require training or education programs focused on skills and productivity.

Key Arguments: Older workers have substantially lower job-finding probabilities than younger workers, and in some contexts over half may become long-term unemployed. The problem is not one cause but a combination of institutional incentives, generous benefits tied to prior earnings, and employer bias toward younger workers with longer career horizons. Policymakers focus less on older job seekers because labor-market policy for them often shifts toward retirement and health rather than re-employment. STEP aimed to update older workers’ job-search methods and re-activate their social networks using intensive group and individual coaching. Random assignment allowed a credible causal estimate of the program’s effects on job-finding and cost-effectiveness. The program increased employment exits by about 4 percentage points from a 40% baseline among controls, implying around a 10% relative increase in job finding. Earnings rose only because people found work faster; the gain in earnings was roughly offset by lower unemployment benefits, so there was little net income effect for participants. From a public-finance perspective, the reduction in benefit payments outweighed the cost of the program, making it cost-effective. The intervention worked better for participants with higher education or higher employability, suggesting that job-search assistance is not equally effective for all older unemployed workers. The program appeared to improve job quality by helping participants obtain permanent contracts and slightly longer hours. Trainer quality mattered, but the study could not identify simple observable traits—such as gender, age, or experience—that explained the differences. The program may be most suitable for recession-driven frictional unemployment, whereas AI-related displacement likely needs different policies centered on retraining and skills development. As retirement ages rise and populations age, helping older workers re-enter work becomes more important for public finances and labor-market participation overall.

Data Points: Age group studied: Over 50 - Eligible unemployed workers targeted by the STEP program Program intensity: 10 group meetings and 2 individual meetings - Description of STEP’s coaching structure Participation rate among invited treatment group: Almost 60% - Share of invited workers who actually participated Participation rate in control group: About 5% - Workers who found some other way to access the program Control-group job-finding / exit rate after about a year: About 40% - Share who found a new job or left unemployment insurance Treatment effect on job finding: +4 percentage points - Increase in the treatment group relative to control Relative increase in job finding: About 10% - Derived from the 4-point gain on a 40% baseline Time frame of evaluation: Around 2015 - Period when the randomized experiment took place Program effectiveness across trainers: Significant differences - Outcome varied by assigned trainer, though no observable trainer trait explained it Job quality outcome: Slightly more working hours and more permanent contracts - Improvement associated with program participation Target population unemployment risk: Over 50% might become long-term unemployed - Describing the risk for older unemployed workers in the studied period

Pivotal Quotes: "Once you have worked for a long time, you climb the job ladder... When you lose your job, you fall off." — Tim Phillips: Introductory framing of why job loss is especially disruptive later in life "The program is cost-effective for the unemployment insurance agency and the government." — Boss van der Klaauw: Summary of the fiscal evaluation of STEP "If you aim for those kinds of changes, you need different programs, you need more training or other kind of educational type of programs rather than job search assistance programs." — Boss van der Klaauw: Explaining why AI-driven labor-market change requires different policy tools than a recession-era job-search program

Implications: Active labor-market policies can help older unemployed workers, especially in recessions, but they are not one-size-fits-all. For structural shifts like AI, policymakers likely need retraining and education, not just job-search coaching.