I document that in the US most of prime-age unemployment is accounted for by relatively few people. I show that their lifetime labor market outcomes are compatible with a model featuring substantial unobserved heterogeneity across workers, in which agents learn workers' types from labor market histories.

We develop a dynamic quantitative model of occupational choice and search frictions with multiple channels of intergenerational transmission (comparative advantage, social contacts and preferences), and use it to decompose the occupational persistence observed in the UK. In the model, workers who choose their father’s occupation find jobs faster and earn lower wages, which is consistent with patterns found in UK data. Quantitatively, parental networks account for 78% of total persistence. Shutting down parental networks or the transmission of preferences improves the allocation of workers and thus yields welfare gains, while removing the transmission of comparative advantage generates welfare losses.

Working papers

What are the returns in terms of children's skills development to child allowance policies? Answering this question requires a theory of the tradeoffs faced by households, as well as a realistic technology of skills formation. I build a model of parental choices which embeds the technology of cognitive and noncognitive skills formation estimated by Cunha et al. (2010), featuring risky investment in children, time use tradeoffs, idiosyncratic income risk and borrowing constraints. Accounting for noncognitive skills implies higher effectiveness of parental investments, and therefore higher policy returns than previously estimated in the literature.

We use administrative linked employer-employee data from Brazil to document that a large share of the gender pay gap is explained by women working at relatively low-paying employers. To shed light on the gender pay gap across employers, we establish three novel facts on revealed-preference ranks of employers by gender. First, employment is concentrated among high-ranked but not necessarily high-paying employers for both genders. Second, women’s ranking of employers is less increasing in pay compared to that of men. Third, both pay and ranks of a given employer differ between women and men. To rationalize these facts, we develop an empirical equilibrium search model featuring endogenous gender differences in pay, amenities, and recruiting intensities across employers. The estimated model suggests that compensating differentials explain one-fifth of the gender gap, that there are significant output and welfare gains from eliminating gender differences, and that an equal-pay policy fails to close the gender pay gap.

We study how employment prospects are related to the employment status of parents using monthly job histories from the UK. We find that individuals whose father is employed have a 8 percentage points higher probability of being employed, driven by a 50 percent higher job-finding probability. These differences are robust to the inclusion of individual fixed effects, and are larger at young ages and when the offspring finds a job in the father’s occupation. Instead, we do not find significant differences by the employment status of mothers. By using a simple model of labor market transitions and on-the-job human capital accumulation, we show that the estimated job-finding probability differentials can account for up to 7 percent of the intergenerational persistence of earnings.

Work in progress

  • The Misallocation of Labor Across Generations (joint with Salvatore Lo Bello) <draft coming soon>