Introduction

The growth in female labour market participation in the post-war decades has raised important questions concerning changes in the factors fostering or inhibiting women's labour supply over the lifecourse. Has the share of women pursuing a continuous or discontinuous career changed? Who, once they have started to work, tend more to exit from and return to paid work? Have these risks changed across generations, and if so, how? In particular, has the trade-off between work and family changed? How and for whom?

In this chapter I first look at the entire observed work trajectory from first job up until the age of 35 in order to furnish a descriptive picture of how many and which women, in each cohort, have entered employment and experienced none, one or several family-care breaks. I shall then focus on specific transitions within the entire trajectory and, using the technique of event history analysis, examine their changing correlates. In other words, as described in Chapter Four, I shall study transitions between employment and housework and the changing effects of a woman's education, labour market experience, occupational class, type of job (full time or part time), marital and childrearing histories, her mother's work experience and the yearly unemployment rate of the region in which she lives. By running first a single additive model for all the four birth cohorts, and then separate models by cohort, I shall also indirectly address the ‘compositional’ issue of the post-war increase in women's employment. The last section summarises and discusses the main findings.

Cohorts, motherhood and types of work history: descriptive evidence

There are various ways to define and identify types of work histories. One of them is to ‘let the data speak’ by detecting typical career pathways using the sequence analysis method (Abbott and Hrycak, 1995; Chan, 1995; Halpin and Chan, 1998; Han and Moen, 2001). Another way is to define significant types of career path a priori by referring to theory and previous empirical findings. Given that here the main concern is to analyse the correlates of different transitions rather than entire trajectories, I shall use the latter method.