From Fixing People to Fitting Work: Seven Theses for Work Inclusion

20 November 20268 min read
Share this article

InkludX Evidence Review, Paper 3 of 3. Count jobs, not exits; make place-then-train the default; build for young and neurodivergent people; separate retention from recruitment; tackle discrimination; redesign the job; spend protection differently, not less.

InkludX Evidence Review on Work Inclusion · Paper 3 of 3 · November 2026

In brief

Papers 1 and 2 reach the same diagnosis. Europe's work-inclusion systems are good at assessing people, protecting their income and, increasingly, moving them off benefits. They are poor at fitting work to people. This paper sets out seven theses for the next decade of work-inclusion policy and practice. Each one is grounded in the evidence reviewed and labelled with how firmly it rests on it.


Why the old paradigm has run its course

For three decades, European policy has swung between two ideas.

  • Protection. Generous benefits for people who cannot work.
  • Activation. Conditions, assessments and programmes to move them back.

Both ideas locate the problem in the individual: their health, motivation or skills. The fix, in both, is applied to the person, and the workplace is treated as given.

The evidence no longer supports that framing.

  • The supply side. The interventions with the strongest evidence put people into real jobs first and support them there. Programmes that try to make people "job-ready" first do worse (Paper 2, Section 1).
  • The demand side. Field experiments show that equally qualified disabled applicants are turned away more often. In a large UK experiment, better references and skills did not change that (Paper 2, Section 4).
  • The design of work. In the United States, the spread of remote work explains most (68–85%) of the post-pandemic rise in full-time employment among people with physical disabilities. That gain came from a change in how work is organised, not from a programme (Paper 2, Section 5).
  • The trends. The sharpest breaks follow changes in benefit rules more often than changes in health (Paper 1, Section 3).

The common thread is that inclusion is produced where people and work meet. Policies that act on only one side get only part of the effect.


Seven theses

1. Count jobs, not exits

Evidence: moderate–strong

The test of a work-inclusion system is how many disabled people are in work, how much they earn and how long they stay. The number of benefit recipients is a different measure.

  • Sweden cut the share of working-age people on sickness compensation dramatically, but disabled Swedes remain far less likely to be employed.
  • Hungary. After a reassessment reform, over a third of people who lost disability benefit ended up with neither work nor benefits.

Every reform should report the employment gap, its effect on poverty, and where those who leave benefits end up. EU-SILC and national registers already make this possible.

2. Make "place, then train" the default — and keep testing it

Evidence: strong for severe mental illness; open for common mental disorders

Individual, job-first support with continued follow-up in the job should be the standard offer. It should not be a pilot project.

  • Where the evidence is strongest: severe mental illness. It is promising for young people with a weak attachment to work.
  • Where it is weakest: the fastest-growing group, people with common mental disorders. Trials for this group often find small effects or none.

That is where the next generation of randomised trials should go. Integrated health and employment services for anxiety, depression and stress-related conditions should be tested against strong usual care, starting within the sick-leave period and running alongside it.

3. Build for young people and neurodivergent minds

Evidence: strong on the trend in most countries reviewed (Germany is an exception); developing on solutions

Exclusion is getting younger, and more of it is linked to developmental, neurodevelopmental and mental-health diagnoses. Disability insurance was largely built for people with a long work history who lose work capacity later in life.

The growing group today has little work history, a diagnosis that can be stable and compatible with work, and a long life ahead. For them, the most expensive decision a system can make is an early permanent award.

We argue for three things:

  • No permanent award without real work experience first. No young person should receive a permanent award without first being offered ordinary work with support.
  • Benefits that combine with part-time work. A route that does not force a choice between benefits and work.
  • Assessment of what a person can do. Assessment that starts from work capacity and job fit rather than from diagnosis.

4. Separate retention from recruitment

Evidence: moderate–strong

The Dutch experience is clear on one side. Making employers responsible for their own sick employees keeps people in work and sharply reduces the flow onto benefits. It may also make employers warier of hiring someone who looks like a health risk.

A system that relies only on employer responsibility protects people who already have jobs and leaves outsiders behind. Recruitment needs its own tools:

  • Wage subsidies aimed at people who are not yet employed.
  • Job intermediaries who carry part of the employer's risk.
  • A guaranteed offer of support to employers, available from the first day of employment.

5. Take discrimination as seriously as we take motivation

Evidence: moderate–strong on the problem; thin on remedies

The gap at the application stage is measured more directly than almost anything else in this field, through controlled experiments. It is remedied less than almost anything else. Disabled applicants face fewer callbacks in nearly every field experiment, across many countries. In the one large test, adding references and extra skills did not close the gap; it widened it.

We propose four remedies:

  • Regular, published correspondence testing, run by public agencies as routine monitoring rather than one-off research.
  • Quotas only where the levy bites, paired with practical help for employers so that complying is easier than not complying.
  • Independent bias audits of AI recruitment tools. The EU AI Act already classifies these tools as high-risk, and audits should cover disability explicitly.
  • Honest employer messaging. It should rest on fiscal and social returns, and on the low cost of most accommodations. It should not rest on correlational claims about profit.

6. Redesign the job

Evidence: promising; mostly from outside Europe

When German employers say there are "no suitable tasks", they are describing how work is organised. The US experience with remote work shows that changing the design of work can move disability employment substantially, without any programme.

Europe should make job design a policy instrument in its own right:

  • A right to request flexible and remote work, with an obligation on the employer to give reasons for a refusal.
  • Support for task redesign and job carving, which means building a role from tasks that suit a particular person.
  • Funding for workplace accommodation that follows the person from job to job.
  • Research that measures what this does to employment, not just to satisfaction. This is the largest gap in Europe's evidence base.

7. Spend protection differently, not less

Evidence: strong on protection; moderate on returns

Social transfers, pensions included, cut the risk of poverty among disabled Europeans by more than two-thirds. That is an achievement, not a failure.

The fiscal pressure is real. In Norway's finance ministry's comparison, fewer transitions to disability benefit gave the largest boost to fiscal room in 2033 of any policy option. The answer is not to withdraw protection. It is to attach work to it: partial benefits that combine with partial work, incentives that make every extra hour pay, and individual support that starts before a permanent award, not after.

The returns are documented. In Norway, IPS for young adult benefit recipients (aged 18–40) is estimated to return more to society, and to the public purse, than it costs.


A research agenda

The gaps in the evidence matter as much as the findings. In order of priority:

  1. Randomised trials of integrated health and employment support for common mental disorders, starting early in sick leave and compared with strong usual care.
  2. Trials of support for young people with neurodevelopmental conditions at the transition from school to work.
  3. Causal evaluations of job design: remote work, flexible hours, task redesign. European evidence on how these affect employment barely exists.
  4. Routine correspondence testing of hiring discrimination in every country, repeated over time.
  5. Studies of hiring, not just retention, after employer-responsibility reforms.
  6. Outcome reporting that tracks where people go when they leave benefits.

Closing

The question for the next decade is no longer whether people with disabilities and health conditions can work. The evidence shows that many can and want to. The question is whether we are willing to change work itself, and the institutions around it, as readily as we have asked people to change.

Europe has built some of the world's most generous and sophisticated welfare systems. Their next task is to become equally good at producing work.


This paper draws on the sources cited in Papers 1 and 2 of the InkludX Evidence Review on Work Inclusion (November 2026). The theses are the authors' own interpretation of that evidence. Where we go beyond what has been tested, we say so.

Tags

evidence-reviewpolicyjob-designinclusion-strategy
Share this article