26
Michael Fertig and Christoph M. Schmidt
Table 3
Probability to be Employed
|
Probit Estimation Results - Dependent Variable: Employed (yes/no) | ||||
|
Variable |
Complete sample |
Restricted sample-Men | ||
|
marginal effect |
t-value |
marginal effect |
t-value | |
|
Individual characteristics | ||||
|
Age (at 24)1 |
0.1175 |
39.29 |
0.1391 |
30.99 |
|
Age (at 40)1 |
0.0105 |
9.63 |
0.0122 |
6.99 |
|
Age (at 60)1 |
-0.1468 |
-30.85 |
-0.1605 |
-24.41 |
|
Female |
-0.2690 |
-80.80 |
- |
- |
|
Medium education |
-0.1294 |
-24.04 |
-0.0702 |
-10.47 |
|
Low Education |
-0.1782 |
-35.45 |
-0.0509 |
-8.14 |
|
Disability |
-0.1721 |
-38.21 |
-0.2287 |
-39.19 |
|
Variables measuring demographic change | ||||
|
Relative own cohort size |
0.0410 |
5.80 |
0.0382 |
3.55 |
|
Share of highly educated |
0.0022 |
8.16 |
0.1925 |
6.08 |
|
Country indicators (base category is UK) | ||||
|
Germany |
0.0923 |
6.77 |
0.0739 |
4.79 |
|
Denmark |
0.1678 |
12.42 |
0.0946 |
6.06 |
|
The Netherlands |
0.0919 |
5.71 |
0.0885 |
5.07 |
|
Belgium |
-0.0207 |
-1.59 |
-0.0198 |
-1.22 |
|
France |
0.0071 |
0.55 |
-0.0244 |
-1.57 |
|
Ireland |
0.0543 |
3.62 |
0.0818 |
5.05 |
|
Italy |
-0.0436 |
-2.58 |
-0.0214 |
-1.07 |
|
Greece |
0.0022 |
0.14 |
0.0689 |
4.08 |
|
Spain |
-0.0730 |
-5.26 |
-0.0090 |
-0.55 |
|
Portugal |
0.1639 |
10.86 |
0.1215 |
7.45 |
|
Austria |
0.1187 |
7.18 |
0.1033 |
5.89 |
|
Finland |
0.0960 |
7.91 |
0.0408 |
2.81 |
|
Sweden |
0.1415 |
11.49 |
0.0714 |
4.98 |
Diagnostics:
|
Pseudo R-squared 0.21 0.26 Number of observations 98,568 48,172 | |
|
1Specification contains the variable and its square. - For a description of the variables see |
RWI |
|
Table 4 (Appendix). |
ESSEN |
For each of our explanatory variables we report marginal effects and their as-
sociated t-values. The marginal effects can be interpreted straightforwardly as
the percentage-point change in the probability to be regularly employed in re-
sponse to a unit-change in the respective explanatory variable, holding all
other explanatory variables, including indicator variables, at their mean val-
ues4. The associated t-values provide an assessment of the statistical reliability
4 For the continuous regressors, such as cohort size or age, a unit-change is a percentage point or a
year, respectively. For indicator variables, the corresponding unit-change is the hypothetical
switch from zero to one.