24(30)
7.2
Empirical modelling and results
The first (basic) model we estimate is the following:
(3)
where
is the annual average number of days absent due to sickness by
child
in family
.
is a dummy variable that takes the value 1 if the child
is the firstborn in the family. This model generates the average difference
between the firstborn and the later-born children of the same gender.
The second model we estimate is:
(4)
where we have added family fixed effects captured by the parameter
.
That is, we also control for joint family components explaining sickness
absence, implying that we explore the variation between biological siblings.
In this estimation we, hence, control for unobserved heterogeneity across
parents, for example, the number of children within the family and the
parents’ sickness absence level.
Finally, the last model we estimate is:
(5)
where we interact the firstborn dummy with a parent dummy that takes the
value 1 if parent
in family
has sickness absence and 0 otherwise. The
term
now captures whether there is a difference in the sibling difference
depending on whether the parents have been absent due to sickness or
not. In all the models, we control for the children’s age and age squared.
The results are presented in Table 8. From the estimates generated by the
first model (column 1), we learn that firstborn children report less sickness
absence than their younger siblings of the same gender. When we control
for family fixed effects (in the second model), the transmission is somewhat
smaller, but still significant. From the estimate presented in column 3, we
can learn that the gap in sickness absence depending on birth order only
emerges in the group of daughters with parents who lack sickness absence.
For boys, the corresponding effect is not statistically significant. Thus, the
birth-order effect seems only to be of importance among daughters and in
families without sickness absence. This result suggests that the higher
intergenerational transmission from parents with higher levels of sickness
absence is driven to a larger extent by joint family components than the
corresponding lower transmission from parents with lower levels of sickness
absence to whom social rank during childhood seems to matters.
