Caseworkers Attitudes

15 / 22

15(21)
Studying only full-time return to work does not alter the results. Neither
does allowing only more permanent (more than one month) returns
to work, ignoring short-term interruptions of the sick-leave episode.
Also, testing alternative model specifications, a Cox proportional hazard
model produces impacts qualitatively the same as the ones generated
by the linear probability model. This also holds allowing for unobserved
heterogeneity in a proportional hazard model with random effects.10
Using survival analyses, the effects can be translated into sickness days.
Comparing the 25 percent most positive caseworkers with the 25 percent
most negative, a positive attitude towards rehabilitation corresponds to
three more sickness absence days on average.11 With an average sick-spell
length of approximately 120 days, this translates into an impact of 2.5
percent. A positive attitude towards the SI rules corresponds to about 3.5
fewer sickness absence days on average, or a 3 percent reduction of the
sick-spell length.
Table 4. Estimates of marginal impact (linear probability model) of
caseworkers´ attitudes on return to work at different durations
Months Attitude (1) (2) (3) (4)
3
Rehab. -0.000
(0.010)
0.004
(0.011)
-0.001
(0.011)
-0.000
(0.011)
SI rules 0.031***
(0.010)
0.030***
(0.010)
0.030***
(0.010)
0.030***
(0.010)
6
Rehab. -0.022**
(0.009)
-0.022**
(0.009)
-0.020**
(0.009)
-0.019**
(0.009)
SI rules 0.029***
(0.008)
0.026***
(0.008)
0.026***
(0.008)
0.027***
(0.008)
9a
Rehab. -0.035**
(0.008)
-0.036**
(0.009)
-0.033***
(0.009)
-0.033***
(0.009)
SI rules 0.024***
(0.007)
0.022***
(0.008)
0.023***
(0.008)
0.022***
(0.008)
Controls
Local office
(LO )
- X X X
Individual
characteristics
(IND)
- - X X
Caseworker
characteristics
(CW )
- - - X
Note: Results from estimations of linear probability models. Model 1 is estimated without any
controls. Model 2 includes local office fixed effects, and Model 3 adds information about the
individual regarding gender, age, educational level, foreign birth, working sector, diagnosis,
children under 18 (yes/no), quarter of sick-leave start, SI benefit, previous days of sickness
absence, unemployment, and disability benefit since 2000. Finally, Model 4 adds information
about the caseworker regarding gender, age, tenure at the SIA, time in the current position,
education, and educational level. We used 65,162 observations. a: The sample only includes sick
spells started in 2010 and 49,676 observations. Standard errors are in parentheses. */**/***
indicates statistical significance at 10/5/1-percent level respectively.
10 See Vaida and Xu (2000) for details of the random effects model.
11 The estimation was performed with the Kaplan Meier-method. We calculate the effects by
summing up the area between the survival curves corresponding to the individuals assigned
the 25 percent most positive and the 25 percent most negative caseworkers with regard to
attitude towards rehabilitation and the SI rules respectively.