Research Interests

My research interests are in labor economics and applied econometrics.


Flexible Scheduling and the Gender Wage Gap


The use of flexible scheduling (allowing the worker control of the start and end times of their work) has been identified as an effective aid for parents balancing work with care giving, and its incidence has been increasing rapidly in Britain and the U.S. at the same time that the rate of convergence in gender pay equality has been slowing. In this paper, I use the 2004 Workplace Employment Relations Survey, a matched employee-employer dataset from Britain, to investigate whether flexibility has positive or negative effects on overall gender pay equality. I find that while men and women have similar access to flexible scheduling on average, the correlates of flexible scheduling differ in significant ways between gender, with family status being more important for women and skill being more important for men. Flexible scheduling is also associated with a higher wage premium for men (6%) than women (2%) after controlling for a large number of worker and firm characteristics, including worker authority and firm fixed-effects. Looking at groups who might experience the largest productivity gains from flexibility, I find that the return to flexibility for women is highest when the woman has young children, but the returns to flexibility for men do not depend on the presence of children. In addition, high-skilled workers of both sexes experience a premium for flexibility, as do workers in jobs that use computers and email. The differences in returns by gender has a large impact on the gender wage gap: for the average worker, flexibility increases the gap by 10%.

Is There a Mommy Track? Occupational Skill and Childbearing


It is well established that mothers earn less than childless women, even after controlling for differences human capital. This paper explores the role of occupational changes after childbirth in contributing to differences in occupational skills and wages between workers of differing family status. The key contribution is to be able to be able to quantify the skill levels needed for each occupation, which allows us to understand when an occupational change involves ``downgrading'' skill. Using data on occupations from the Dictionary of Occupational Titles and the O*Net combined with panel data from the National Longitudinal Study of Youth (1982-2000), we find little change in math skills for new fathers but sizable decreases in math skills (as well as language and reasoning skills) for women, particularly women with two or more children. This occupational ``downgrading'' is the largest for highly educated women with more than a high school degree, who on average move 7 to 10 centiles down in the overall distribution of female math skills after having their second and third children, respectively. This translates into lower wages as well, accounting for up to 10% of the motherhood wage gap for highly educated women and up to 4% for women with lower levels of education. We find that changes in occupational attributes such as the pleasantness of the work environment, job stress, and part-time work explain the bulk of the skill downgrading for women, suggesting that the bundling of family-friendly job attributes with less skilled work is one cause of the the observed ``mommy-track'' for women with multiple children.

Flexible Work Arrangements and Wages: Do Firms Matter?


Using data from the 2004 Workplace Employment Relations Survey, the author examines whether the relationship between flexible work arrangements and wages depends on which types of firms choose to offer flexibility. Past studies have found a positive relationship between flexibility and wages, but have not been able to control for management practices that may be positively related to both wages and the decision to offer workers flexibility. The results suggest that the firm matters for all types of flexibility considered, and that the correlation between wages and flexibility is much smaller when holding the firm's environment constant. The author also finds evidence of compensating differentials for women, particularly in industries that have low availability of flexible work arrangements.

Revisiting the Motherhood Wage Penalty


This paper investigates the motherhood wage penalty, or the unexplained portion of the wage gap between mothers and non-mothers after controlling for observable characteristics. The size of this penalty has been of great interest to economists because it may represent employer discrimination against mothers or lower productivity because of family responsibilities. However, rather than a truly causal effect, the penalty could be due to the presence of unobserved heterogeneity, endogeneity, or sample selection that bias OLS estimates. To investigate this possibility, I apply the fixed effects estimator with instrumental variables (FE/IV) to panel data from the NLSY for the years 1988-1998. I conclude that the wage penalty does not represent a causal relationship between children and wages. This finding implies that the attention of researchers and policy makers should be focused on differences in the observable attributes of mothers relative to childless women (and the return to these attributes) in explaining the motherhood wage gap, particularly differences in work experience and education.

The Effect of Head Start on Maternal Employment:An Evaluation of Non-Experimental Estimators of Treatment Effects


This research investigates whether government-provided child care increases the employment of low-income mothers. I use NLSY Head Start enrollment data to calculate non-experimental estimators of the average treatment effect of participation on the mother's employment, including matching and weighting on the propensity score. I find no statistically significant effect of the treatment on employment using these methods, a finding confirmed by using several comparison groups. In addition, negative and significant effects are found for white mothers. However, using a regression discontinuity (RD) design resulted in small positive effects of Head Start participation for mothers' employment growth (5\%) for some sample restrictions, but no effect using other samples. For those mothers participating in welfare, some sample restrictions using RD resulted in a larger positive effect of Head Start of 8\%. The RD estimates differ substantially from those of the matching and weighting estimators, which could suggest that the latter do not fully remove the selection bias.