Sweden: From Disability Pension to Work-First — A Policy Experiment in Progress

The Swedish Reform Experiment

In 2008, the Reinfeldt government enacted one of the most radical disability benefit reforms in OECD history. The sjukersättning (sickness compensation, formerly known as the disability pension) was frozen to new entrants, time-limits were placed on sjukpenning (sickness benefit), and a new mandatory "rehabilitation chain" (rehabiliteringskedja) imposed strict return-to-work assessments at fixed intervals.

The caseload results were dramatic. The number of people on long-term sickness compensation fell from approximately 570,000 in 2006 to under 280,000 by 2012 — a reduction of more than 50% in six years.

Swedish and international observers initially described this as a remarkable policy success. The story was more complicated.

Where Did the Caseload Go?

Three major research projects have tracked the trajectories of people who left the sickness compensation system post-2008:

Stockholm University (Stenberg & Westerlund, 2015): Tracking 44,000 individuals who were terminated from sickness benefit under the new rules, the study found:

• 28% entered regular employment within 24 months
• 31% entered unemployment benefit (A-kassa)
• 19% entered social assistance (försörjningsstöd)
• 14% entered other transfers (student benefit, housing allowance)
• 8% had no measurable income source in administrative data

The employment outcome — 28% — was substantially lower than reform proponents had projected (40–50%).

IFAU (Institute for Labour Market and Education Policy, Johansson et al., 2018): Using a regression discontinuity design exploiting the age threshold in the reform, IFAU found that the reform increased employment by approximately 5 percentage points among those affected — significant, but concentrated among those closest to the labour market (low health impairment, recent work history). For those with severe conditions, the primary effect was transfer to unemployment benefit or social assistance, not employment.

Inspektionen för socialförsäkringen (ISF, 2020): A comprehensive audit found that approximately 40% of individuals terminated from sickness benefit had been regranted benefit within 36 months — either through new sick-leave spells or through successful appeals. This "revolving door" effect substantially erodes the apparent caseload reduction.

The Social Insurance Assessment Controversy

The Försäkringskassan (Social Insurance Agency) assessment process became the subject of intense political controversy between 2015 and 2020. Parliamentary investigations documented:

• Assessment rejections at rates significantly above medical evidence — the government ombudsman found that Försäkringskassan had applied an overly strict "work capability" standard that did not adequately account for co-morbid conditions
• Regional variation in rejection rates of up to 30 percentage points for identical clinical presentations
• A systemic backlog: by 2019, over 50,000 appeal cases were pending before the administrative courts

The Löfven Social Democrat government responded with a partial reversal — loosening assessment criteria and temporarily halting the mandatory rehabilitation chain for cases with serious psychiatric diagnoses. The number of new sickness compensation grants increased from 2019 onwards.

What Works in Sweden: The Evidence Base

Swedish labour market policy has generated rigorous evidence on several intervention types:

Samordningsförbund (Coordination Associations): Multi-agency coordination bodies bringing together Försäkringskassan, Arbetsförmedlingen (Employment Service), municipalities, and healthcare. A 2019 evaluation by Nationella rådet (NORA) found coordination association participants showed 15–22% higher rates of employment entry compared to matched controls receiving standard services — with the strongest effects for those with complex, multi-barrier profiles.

Supported Employment (SE-model): Sweden's version of IPS, implemented through Arbetsförmedlingen's Supported Employment programme. A 2021 quasi-experimental study by Uppsala University found SE participants achieved competitive employment at 41% vs 19% in traditional vocational rehabilitation — consistent with the international IPS literature.

Early intervention for mental health: The PAVES programme (Psykiatrisk Arbetsrehabilitering Vetenskapligt Utvärderad i Sverige) tested integrated psychiatric and vocational support. The RCT (Lagerveld et al., 2021 — cross-Nordic study including Sweden) found 29% of IPS-plus-CBT participants were in competitive employment at 24 months, vs 14% in treatment-as-usual.

The Current Picture

As of 2024, Sweden has approximately:

• 188,000 on sjukersättning (sickness compensation/disability pension)
• 90,000 on long-term sjukpenning (sickness benefit, 180+ days)
• 110,000 on activity compensation (aktivitetsersättning, under-30 cohort)
• 380,000 on unemployment benefit or job-seeker activity programmes

InkludX estimates direct benefit expenditure for the working-age excluded population at approximately €8.8 billion annually, rising to a total fiscal burden (including lost tax revenues) of €13.3 billion, or 2.3% of GDP.

The Political Balance Sheet

Sweden's 2008 reform demonstrated that aggressive tightening of benefit criteria can reduce recorded caseloads rapidly. It also demonstrated the limits of a "work-first" approach applied without adequate labour demand, employer support, or workplace accommodation infrastructure.

The lesson most frequently drawn by comparative policy analysts (OECD, 2023; Anxo & Niklasson, 2022): activation without accommodation produces churn, not employment. The workers who most need labour market support — those with severe mental health conditions, multiple disabilities, long benefit histories — are precisely those for whom a supply-side shock to benefit conditionality is least effective.

Sweden's policy challenge for the 2020s is to build the demand-side infrastructure — supported employment at scale, employer incentive reform, workplace health services — that the 2008 reform assumed would follow automatically but did not.

---

Sources: Statistics Sweden (SCB) 2024; Försäkringskassan Statistik 2024; IFAU Rapport 2018:15; ISF Rapport 2020:8; Stenberg & Westerlund, Stockholm University 2015; OECD Employment Outlook 2023.