306
Journal of Applied Economics
Several features are worth mentioning in the sample.9 First, as already shown,
women are far more likely to be employed in the public sector. Women in the public
sector are also significantly less likely to have children aged 0-11 but slightly more
likely to have older children compared to men. This also holds for the private
sector.
Second, union coverage is significantly higher in the public sector for both
men and women. This also carries over into the perception of unions. Employees
of both sexes in the public sector perceive unions as an important institution.
Furthermore, public sector workers are more likely to be married and exhibit far
greater job attachment. Yet, there are only minor differences in terms of educational
levels. On the other hand, as already mentioned, both men and women in the
public sector are more likely to be found in managerial positions compared to their
private sector counterparts.
B. Estimation results
Univariate and bivariate results for both males and females do not support a
simultaneous estimation of the participation and sector choice decision.10 The
correlation coefficient ρuv is not significantly different from zero. Furthermore, the
assumption of correlation does not change the estimated coefficients substantially,
which is not uncommon in this literature (e.g., Fishe et al. 1981 and Tunali 1986).
For male employees the following patterns with respect to participation and
sector choice arise. First, young people (aged 16-20) are significantly less likely to
both select themselves into the public sector and participation compared to the
base category (employees aged 31 and older). Many studies have ascribed the age
effect to queuing for public sector jobs (see, e.g., Bender 1998 and van der Gaag
and Vijverberg 1988).
Second, married men are significantly more likely to participate in the labour
market compared to unmarried males. Yet, the marital status has no impact on the
sector choice. Similarly, men with children (aged 0-11) are more likely to be found
working compared to males without children. This holds also for men with older
children, however, the effect is not statistically different from zero in the univariate
probit case.
Third, the identification variables on trade union perception perform well.
9 A summary table with sample characteristics can be obtained upon request.
10 Full tables with univariate and bivariate results can be obtained from the author.