Characterizing Leave for Maternity: Modeling the NLS-Y Data

Major changes in women's labor force behavior over the last two decades imply that while time away from the workforce after the birth of a child was once measured in years, it is now measured in weeks or even days. Concentrating on the weeks immediately following childbirth, this paper characterizes the labor force behavior of women immediately before and after the birth of a child. The timing of labor market exits (during pregnancy) and entrances (after childbirth) are estimated to the day, and reported to the week. Quits, exits to unpaid leave, and exits to paid leave are separately identified. The estimates reveal that most women who work before the birth of a child return to work relatively quickly after the birth of a child. The model time to return occurs only about six weeks after childbirth. Those who work long into pregnancy return to work more quickly after childbirth. The empirical work uses the National Longitudinal Survey-Youth (NLS-Y). The estimates are generated using a system of profit and hazard models. The system includes unobserved heterogeneity to capture the correlation between decisions. The econometric model is specified to correct for the focus of the NLS-Y protocol (in some years) on employment, so that it is not possible to distinguish paid from unpaid leave.