11(23)
Table 2 reports sick-spell descriptives. A comparison suggests that the
difference in sick-spell length between the spells started in June and July
each year has become somewhat smaller, from -3.0 (56.7-59.7) to -1.2
(56.1-57.3) days. Hence, sick spells initiated in June became shorter
between 2007 and 2008 – in correspondence with the downward trend in
sick-spell length – but the July spells became even shorter on average
during the same period.
The sickness-spell register lacks information on the cause of ending a
sickness period. We would however assume that the vast majority of the
individuals ending a sick spell within 6 months returned to their current
employer. By analysing the most common alternatives to returning to work
– transitions to unemployment and disability benefits – we investigate how
good an approximation return to work is for all exits from sickness absence.
From Table 2, we note that only 0.3-0.8% of the sample ended the sick
period to become unemployed. Even fewer (0.0-0.1%) were granted
disability benefits. These small shares make state-specific analysis
redundant. In the rest of the paper, I use sick-spell exits and return to
work synonymously.
Table 2
Sick-spell descriptives
July (1
st
week)
2008
June (last
week)
2008
July (1
st
week)
2007
June (last
week)
2007
Share of spells at least (days):
30
59.9
57.5
61.1
60.0
90
17.9
17.4
19.1
16.8
180
8.7
8.1
10.2
8.6
Average (days)
a
57.3
56.1
59.7
56.7
Share of spells ending with:
b
Disability benefits
0.0
0.0
0.0
0.1
Unemployment
0.5
0.3
0.8
0.6
Unknown (return to work)
92.1 92.5 90.1 91.6
Ongoing spells (28 weeks)
7.4
7.2
9.1
7.7
Note:
a
Spells still in progress at 196 days have been given the spell length 197.
b
Some observations are lost combining different registers; this modified data set
contains 18 174 observations instead of 19 211. The transitions refer to full-time exits
from sickness payment, in contrast to part-time exits.
