Research

We study how turnover and human capital dynamics shape the life-cycle gender pay gap when employers are forward-looking and able to set gender-specific wage rates. In our equilibrium wage-posting model with learning-by-doing and fertility events, the life-cycle gap can be attributed to worker productivity, job search, employers’ endogenous wage-setting, and job productivity. Estimating the model on NLSY79 data, we find that  although the high school and college gaps are driven by different forces, employers' wage-setting accounts for one-third of the gender gap in both groups. Neglecting interactions between turnover and human capital dynamics biases down the estimated role of turnover substantially.

Women face a "glass ceiling" in career advancement, with considerable underrepresentation in high-paying, high-status positions. This may arise because employers anticipating more career interruptions from women might allocate them to low-productivity jobs where turnover is less costly. This paper develops an equilibrium search model to quantify the extent to which the "glass ceiling" could be attributed to employers' job allocations decisions on one hand, and women's experience accumulation and preference for family-friendly amenities on the other hand. Estimating the model on administrative employer-employee data from Finland, I find that employers' differential wage-setting and job allocations account for a substantial proportion of the gender wage gap in early career, whereas gender differences in labor force attachment explain most of the wage gap in late career. The model reveals that sharing parental leave more equally between men and women closes gender gaps both before and after having children.

We introduce a new concept called partial efficiency (PE) to model the post-divorce behaviors of ex-spouses. We assume that divorced parents still care about their children and maintain an efficient approach to the provision of the public good, but they do not share risk or compensate each other's private consumption. We show that the PE approach offers more realistic implications than the full-efficiency (FE) or non-cooperative (NC) models of divorce.