17(30)
5.2
Regression models
The model we estimate with OLS is the following:
(1)
where
is the annual number of days absent due to sickness of the son
(daughter) in family
and
is the corresponding measure for his (her)
father (mother). This model estimates the intergenerational transmission of
sick leave. We use the average annual sickness absence over the observed
six years throughout and we control for the ages of the fathers (mothers)
and sons (daughters) linearly and quadratically in all of our regressions.
The parameter
captures the marginal magnitude of the transmission in
terms of annual days absent.
In order to obtain easily interpretable estimates we estimate OLS
regression models. To mitigate the large variation in sickness absence
across the population, we control for whether the parent has sickness
absence or not in several model specifications; however, we start by
presenting the average estimated mobility.
5.2.1
Average transmission
Table 4 presents estimates of the average marginal transmission of sick
leave (model 1), separately by the gender of the parent and the child. From
mothers to daughters the interpretation of this estimate is that for every
additional day the mother is absent, the daughter’s absence increases by
0.115 days. If we extrapolate this result, we can say that if a mother
increases her absence by 10 days, her daughter’s absence increases by
1.15 days. Ten days corresponds to about half of the average number of
days absent due to sickness among mothers and just over one day
corresponds to about 4 per cent of the average number of days absent due
to sickness among daughters. (The average days are presented in Table 1.)
Again, if we extrapolate, the interpretation is as follows: a mother with
about 50 per cent more sickness absence than the average mother’s level
has on average a daughter with about 4 per cent more sickness absence
than the average daughter’s level. The corresponding transmission from
parents to daughters is somewhat smaller than the corresponding
transmission to sons, irrespective of the gender of the parent.
Table 4.
Average intergenerational transmission of sickness absence
days
Dependent variable: the child’s average sickness absence days
0–365
Daughter
Son
From mother
0.115***
0.0663***
(0.00238)
(0.00182)
From father
0.0880***
0.0682***
(0.00255)
(0.00195)
Observations
224, 385
224, 385
238, 599
238, 599
R-squared
0.0166
0.0081
0.0106
0.0072
Note: We start counting the number of days absent with sickness benefits from the
fourth week and with disability pension from the first day within a sickness absence
spell. In all the models, we control for both the child’s and the parent’s age and age
squared. Standard errors are in parentheses. Statistical significance: *** p<0.01.
