Ongoing work

This paper examines how skills acquired and reported by students on their resumes contribute to the early-career gender wage gap even among graduates from the same field of study. Using a rich dataset of thousands of complete university students’ resumes matched with the job postings they viewed on the university job board, we find that, on average, women view job postings with salaries 8.6\% lower than those viewed by men and that differences in viewed postings across fields strongly correlate with the gender wage gap reported by Statistics Canada. Applying language processing techniques, we construct a taxonomy of 2,169 skills and use it to decompose the wage gap into a skill effect – as skills more commonly reported by women are associated with lower wages – and a behavioral effect – as women tend to view lower-paying job postings than men with similar skills. The decomposition analysis shows that the skill effect explains 48% of the wage gap when all fields are combined, although this share varies significantly between groups of majors. To better understand the behavioral effect, we focus on business students and find that women are more likely to seek jobs in smaller firms, prefer roles in human-centered industries, and exhibit signs of underconfidence when engaging with job titles, favoring roles such as ‘Assistant’ or ‘Coordinator’ over ‘Analyst’ or ‘Director’.

In this paper, we investigate the link between the presence of various amenities in job postings and the proportion of women in the relevant occupations. We analyze thousands of job postings from a free government job board in the province of Québec, Canada, using a mix of techniques from natural language processing and standard econometrics, and show strong evidence that job postings in men-dominated occupations are more likely to propose financial incentives, such as bonuses or pay based on performance, women-dominated occupations are more likely to contain mentions of various non-monetary incentives, even when controlling for industry or administrative region.

This paper introduces the notion of subjective disability-free life expectancy (SDFLE), defined as the number of years individuals expect to live without health limitations. We decompose expected remaining lifetime into years anticipated without limitations and years anticipated with limitations, providing a direct measure of perceived future exposure to disability. Using French data from the Survey of Health, Ageing and Retirement in Europe (SHARE), we construct individual subjective survival and disability-free survival curves and compare them to objective period and cohort life tables under alternative disability scenarios. We document marked gender differences. Men’s expectations are broadly consistent with projected longevity and disability trends. Women, however, underestimate total longevity while anticipating improvements in healthy years, leading them to underestimate the number of years they are likely to live with disability. These findings inform the use of subjective expectations in demographic projections and long-term care planning.

This project aims to gain a a better understanding of nurses’ prefer-
ences related to their work conditions. In recent years, the Québec
government tried to increase the number of hours worked by nurses
by increasing their salary, only to discover that the wage elasticity
of these workers was very low in the current condition. We use the
stated-preferences approach to determine how various work condi-
tions affect nurses’ wellbeing, with a particular emphasis on manda-
tory overtime and unusual work scheduling.

We analyze survey data including respondents with and without disabilities, and both employed and unemployed. We elicit probabilistic expectations on key labour market outcomes—such as interview invitations, job offers, earning minimum wage, job separation, and job promotions—using a 0–1 scale, allowing for cardinal and inter-personally comparable measures of beliefs. To address the measurement challenges associated with ordinal disability indicators, we implement a flexible, data-driven approach based on the fused LASSO, which exploits the ordered nature of functional limitation categories without imposing arbitrary thresholds. Our results show that the mapping between reported functional limitations and economically meaningful disability groupings varies substantially across domains. We further document significant differences in subjective labour market expectations across disability domains.