5(23)
1
Introduction
In 2003, Sweden had the highest reported sick-leave rate in the EU-15 with
4.4% of the working population being sick absent
A relatively large amount of empirical literature has investigated the
correspondence between the generosity of the SI and sickness absence. For
example, Johansson & Palme (1996, 2002 and 2005) and Hesselius &
Persson (2007) all find evidence of higher compensation levels increasing
the costs in the SI system. The negative incentive effects from a generous
insurance system could partly be offset by monitoring and sanctions. The
effectiveness of these policy instruments is however much less analysed.
On Swedish data, Hesselius, Johansson and Larsson (2005) make use of a
unique large-scaled experiment in 1988. They find strong evidence of more
extensive sick spells as a result of prolonging the medical-certificate-free
period from 7 to 14 days. The result is expected from the theory of moral
hazard predicting higher insurance costs when the insurance conditions
become more generous (see for instance Barr, 2004).
Together with high
inflows to the disability insurance, a large share of the workforce became
dependent on income from the public social insurance systems. Despite a
positive trend in recent years with fewer sick spells starting and shorter
sickness spells, public sickness insurance (SI) underwent a major
reformation in 2008. Several changes were introduced with the purpose of
reducing excess use of the SI and supporting individuals better in their
return to work. The most radical restructuring was the introduction of time-
restricted assessments of working capacity for certain sick-spell durations:
the so-called
rehabilitation chain
.
In the relatively large empirical literature on the unemployment insurance
(UI), monitoring through stricter enforcement and verification of work-
search requirements have generally been shown to reduce benefit periods.
Both the Washington Work-Search Experiment (Johnson & Klepinger, 1994)
and the Maryland UI Work-Search Demonstration (Klepinger et al, 2002)
evaluate the impact of various combinations and degrees of work-search
requirements and monitoring. The analyses provide strong evidence that
such activities increase the UI exit rate. In Europe, Dolton & O’Neill (1996)
find large positive effects of introducing recurrent counseling and
monitoring meetings (the British Restart program) for long-term
unemployed.
No earlier study has investigated the impact of introducing time limits in
the public SI system. The question to be answered in this paper is whether
the introduction of repeated working capacity assessments on the 91st and
181st days of the sickness period reduce the sickness absence length and
increase return to work. To separate the effect of the assessments from the
downward trend in sickness absence, the analysis exploits a quasi-
experimental feature implementing the policy. Sick spells initiated only a
few days apart (in the last week of June and the first week of July) were
treated under different regimes during the first six months. Including spells
started in the same weeks one year earlier, a difference-in-difference
strategy is used to estimate the reform’s impact. Job-search theory predicts
that the notification of future monitoring will have an immediate negative
effect on the individual’s value of remaining absent from work. This is
expected to affect the return to work through increased efforts and a higher
1
Swedish Social Insurance Agency (2009). The EU-15 refers to the members of the
EU in 1995. The average sick-leave rate among these countries was 1.9% in 2003.
