The pdf of her Philadelphia paper is here. This is from the concluding section:
The reasoning of this essay is as follows. A gender gap in earnings exists today that greatly expands with age, to some point, and differs significantly by occupation. The gap is much lower than it had once been and the decline has been largely due to an increase in the productive human capital of women relative to men. Education at all levels increased for women relative to men and the fields that women pursue in college and beyond shifted to the more remunerative and career-oriented ones. Job experience of women also expanded with increased labor force participation. The portion of the difference in earnings by gender that was once due to differences in productive characteristics has largely been eliminated.
What, then, is the cause of the remaining pay gap? Quite simply the gap exists because hours of work in many occupations are worth more when given at particular moments and when the hours are more continuous. That is, in many occupations earnings have a nonlinear relationship with respect to hours. A flexible schedule comes at a high price, particularly in the corporate, finance and legal worlds.
A compensating differentials model explains wage differences by the costs of flexibility. The framework developed here shows why there are higher or lower costs of time flexibility and the underlying causes of nonlinearity of earnings with respect to time worked. Much has to do with the presence of good substitutes for individual workers when there are sufficiently low transactions costs of relaying information. Evidence from O*Net on occupational characteristics demonstrates that certain features of occupations that create time demands and reduce the degree of substitution across workers are associated with larger gender gaps.
Data for MBAs and JDs shows large increases in gender pay gaps with time since degree and also reveals the relationship between the increasing gender pay gap and the desire for time flexibility due to the arrival of children. Lower hours mean lower earnings in a nonlinear fashion. Lower potential earnings, particularly among those with higher-earning spouses, often means lower labor force participation. Pharmacists, on the other hand, have pay that is more linear with respect to hours of work. Female pharmacists with children often work part-time and remain in the labor force rather than exiting.
The paper is interesting throughout.
Addendum: Mary Ann Bronson, a job candidate from UCLA, has a new and interesting paper (pdf) on the gender gap across college majors and related issues. Here is another UCLA job market paper, by Gabriela Rubio, on why arranged marriages decline in frequency. This year, at Duke University, there are more female entering students in the Ph.d. program than male.