The aggregate probability of ending a career interruption that begins at childbirth is shown to diminish rapidly with the length of the interruption. The empirical models estimated suggest that the decline can be explained by a combination of structural duration dependence, unobserved heterogeneity, and differences in observed characteristics. The probability of returning to employment for a group of women with identical observed characteristics is found to have an L-shaped distribution for the majority of the sample, suggesting that many women have a very low probability of returning.

This study provides rationale for the underrepresentation of women in the pension sector and examines the consequences for the gender wage gap. For a given set of observed characteristics, a woman is 11-19 percent less likely than a man to have a pension. Of the unexplained portion of the gender wage gap, 10-38 percent is due to unexplained differences in pension coverage. Finally, consistent with a screening effect of pensions, women are paid more equally in the pension sector.