Gender, Incentives and the Division of Labor

Author: Ylva Moberg, And

Summary of

Dissertation series


Essay 1

The length of parental leave entitlements is known to affect take-up rates, division of parental leave between parents, and the mother's decision to return to work. So far, however, the importance of the level of benefit has received little attention in the literature. Using population wide register data, I exploit the “speed premium” rule in the Swedish parental leave system as a source of random variation in the benefit level. A fuzzy RD strategy is used to estimate the causal effect of a change in the level of benefits per day on the utilization of parental leave among Swedish parents. The results suggest that parents’ take-up of benefits is highly sensitive to the benefit level. A 1% (5 SEK ≈ 0.54 $) increase in the mother's benefits per day is found to increase her length of leave by about 1 % (2.6 days).  This translates into an elasticity of take-up duration (length of spell) with respect to the benefit level of 1, a parameter that has not been estimated before. Fathers respond to the increase in mothers’ take-up by reducing their time on leave by an almost equivalent number of days (1.9 days). In other words, the change in benefit level affects not only the individual’s take-up, but the division of parental leave between parents.

Essay II

In this paper, I compare the effect of entering parenthood in lesbian and heterosexual couples using Swedish population-wide register data. Comparing couples with similar pre-childbirth income gaps, a difference-in-differences strategy is used to estimate the impact of the gender composition of the couple on the spousal income gap after childbirth. The results indicate that the gender of the parents' does matter for their division of labor as, five years after childbirth, the income gap is significantly smaller in lesbian than in heterosexual couples, also when comparing couples with the same pre-parenthood income gap. Part of the explanation is a difference in biological restrictions: lesbian partners often give birth to one child each and spend more time at home with the child they carried. Other explanations are the influence of gender norms and differences in preferences between lesbian and heterosexual couples.

Essay III

The skewed division of parental responsibilities during a child's infancy is often assumed to be a natural consequence of the mother being pregnant and wanting to breastfeed. In this paper, I investigate to what extent the tendency to let the mother be the main caregiver of an infant can be explained by the fact that she is the one to be pregnant, not the father. Using the division of parental leave during the child’s first two years with the parents as a proxy for the division of parental responsibilities, I compare the behavior of biological parents (where the mother gave birth) to adoptive parents (where she did not) in Swedish population-wide register data. My results show that adoptive parents, both mothers and fathers, spend less time on parental leave than biological parents, but that the mother's share of leave is about the same as among biological parents. There is thus some support for the hypothesis that a biological tie increases parents’ initial investment in children, but not that this relationship is stronger for women. Hence, there is no evidence that the mother’s birth giving status can explain her share of parental responsibilities. Due to methodological challenges, it is difficult to disentangle the different mechanisms that could explain the results.

Essay IV

with Spencer Bastani and Håkan Selin

No previous quasi-experimental paper has systematically examined the relationship between the extensive margin labor supply response to taxation and the employment level. We model the labor force participation margin and estimate participation responses for married women in Sweden using population-wide administrative data and a solid identification strategy. The participation elasticity is more than twice as large in the lowest-skill sample (with relatively low employment) as compared with the highest-skill sample (with high employment). Our analysis suggests that cross- and within country comparisons of participation elasticities always should be made with reference to the relevant employment level.