S9 Ep36: Helping the over-50s find work

Lose your job at 25 and someone will help you find another. Lose it at 55 and the talk quietly turns to how you might wind down towards retirement.

Policymakers tend to assume job search training works for the young and not the old, so they rarely spend money trying. Bas van der Klaauw (Tinbergen Institute) thinks they got that wrong.

In this week's VoxTalks Economics, he tells Tim Phillips about a Dutch experiment that put older unemployed workers through an intensive programme built on one idea: teach people over 50 to find work the way younger workers already do, by working their social network.

Participants left unemployment faster, there was a 10% increase in job finding, and the savings in benefits more than covered the cost. The catch: it helped the better educated most and was tested in a recession. Will it work just as well in today's labour market, where even the young and well-educated are struggling to find good jobs?

The research behind this episode:

de Groot, Nynke, and Bas van der Klaauw. 2026. "A Randomized Experiment on Improving Job Search Skills of Older Unemployed Workers." CEPR Discussion Paper 21464. (Gated)

To cite this episode:

Phillips, Tim, and Bas van der Klaauw. 2026. “Helping the over-50s find work.” VoxTalks Economics (podcast).

About the guest

Bas van der Klaauw is professor of economics at Vrije Universiteit Amsterdam and director of the Tinbergen Institute. An applied microeconometrician, he uses causal methods to study labour markets, education and health, and is a research fellow of CEPR and IZA. His work on unemployment insurance, active labour market programmes and job search includes several field experiments run with the Dutch benefits administration.

The paper is co-authored with Nynke de Groot, an economist at the National Health Care Institute (Zorginstituut Nederland) who took her PhD at Vrije Universiteit Amsterdam. Her earlier work with van der Klaauw includes a study of how cutting the unemployment insurance entitlement period affects job finding.

Research and concepts discussed in this episode

Older workers and long-term unemployment. Older unemployed workers tend to have job finding rates around half those of younger workers, and during the recession the study covers, more than half of older job seekers risked becoming long-term unemployed. Van der Klaauw attributes the gap to a combination of factors rather than any single cause: more generous and longer benefit entitlements that weaken the incentive to take a lower-paid job quickly, and employers who favour younger hires expected to grow with the firm over a longer horizon.

STEP (Successfully to Employment Program). A Dutch job search assistance programme developed during the post-2008 recession for unemployed workers aged 50 and above who had not found work within a few months of claiming unemployment insurance. It ran to 10 group sessions of around four hours each plus two individual meetings, covering interview practice, CV writing and social media, with a particular emphasis on activating the participant's social network. Participants were encouraged to have at least one conversation a week with a contact about possible work. The programme cost roughly 470 euros per participant.

The experiment. The study covers everyone aged 50 to 63 who entered unemployment insurance in the Netherlands between November 2014 and July 2015 and remained unemployed for three months, about 50,000 people. Assignment to treatment or control was based on the last digit of the social security number, putting roughly 20% in the control group. Because participation was voluntary (an encouragement design), the authors report both the effect of being offered the programme and, using random assignment as an instrument, the effect of actually taking part. Around 54% of those in the treatment group took up STEP.

What it did to job search behaviour. The job application register lets the authors watch how people searched. Participants made fewer applications to posted vacancies and did more networking, exactly the shift the programme was designed to produce. The change in method did not raise the number of job interviews, but it was accompanied by faster exits from unemployment.

Cost effectiveness. Participation cut cumulative benefit payments by about 715 euros within 18 months, comfortably above the 470 euro cost, making STEP cost effective for the benefits administration. For participants, the lost benefits were almost fully offset by higher earnings from working sooner, so there was no large income gain to the individual, but no loss either.

Who it helped. Effects were strongest for the better educated, those with higher pre-unemployment earnings and those not previously working through a temporary work agency. There was little or no effect on the lowest educated, who also had the lowest take-up. The authors find no significant difference by gender or by age band within the 50 to 63 range.

Trainers and group composition. Effectiveness varied significantly across trainers, but no observed characteristic (gender, age, experience, contract type) explained which trainers did better. Group composition mattered too: participants did better when their group contained some lower-educated members, which argues for mixed rather than streamed training groups. One reading is that trainers may concentrate their attention on the more employable members of a group.

Does it generalise? Two caveats. The programme was evaluated in a recession, when people were losing jobs through no fault of their own (frictional unemployment), and it may do less when work is easier to find. And it was designed for that kind of unemployment, not for the structural problem of workers whose skills no longer match available jobs, where van der Klaauw suggests training or retraining, rather than job search help, is the relevant tool.