Research
Abstract: We investigate gender productivity gaps in science using a unique dataset of over 93 thousand professors and researchers from French higher education and research institutions. This dataset spans nearly 60 years and includes detailed information allowing us to observe age, gender, occupations, career spans, and potential name variations on top of detailed publication outputs. Carefully controlling for composition effects and various fixed effects, we find that not only does the gender gap persist in recent cohorts, but it exhibits a specific pattern throughout the career cycle, consistent across scientific productivity indicators, occupations, cohorts, and fields of science. The gender gap increases in the first part of the career, reaching a maximum of around 42% of age, before slightly decreasing until the end of the career. One interesting exception is life sciences, where the gap never decreases. Using additional observational data, we investigate the role of various contributing factors, such as selection bias, childbirth, teaching loads, promotion incentive biases, team leadership, and funding, to understand the underlying mechanisms driving these trends. We discuss which science policies may robustly mitigate the mechanisms shown to increase gender gaps.
Breaking the Cluster: How Collective Funding Reshapes Gender Gaps in Coauthorship Networks, with Nicolas Carayol and Pascale Roux Funding: Région Nouvelle Aquitaine Grant (AAPR2020-2019-8020710). PI: Pr. Pascale Roux
Abstract: Women's coauthorship networks are smaller, more locally clustered, and less likely to bridge disconnected parts of the scientific community than men's, a pattern often attributed to unequal access to the resources that support tie formation. Testing this resource-constraint hypothesis is difficult because funding under individually-targeted grant schemes is itself shaped by gendered self-selection into competitive environments. We exploit the staggered rollout of the Laboratoires d'Excellence (LabEx) program, a large French cluster-based funding scheme awarded to research communities rather than individuals, to estimate the effect of collectively relaxing resource constraints on gendered network formation. Using a regression-based staggered difference-in-differences estimator on three-year rolling coauthorship networks for 4,224 researchers, and relying on grades given by the selection committee to control for selection into treatment, we find that LabEx funding expands coauthorship networks for both genders, but substantially more for women, and through different channels: men's modest gains are concentrated within the funded cluster, while women's larger gains occur almost entirely outside it, including internationally. Among the most productive researchers this asymmetry sharpens dramatically, with treated women's networks expanding roughly threefold relative to the full sample while equally productive treated men show no outward expansion at all. Both genders gain funded grants post-treatment, but only top-quartile women show a corresponding rise in grant applications, suggesting that institutional certification specifically lowers barriers to competitive grant-seeking for highly productive women. These findings identify constrained network formation as a mechanism linking funding disadvantage to broader gender gaps in academic careers, and show that collective, cluster-based funding instruments can partially offset it.
A Portrait of Gender in French Mathematics: Representation, Specialization, and Productivity, single-authored work
Summary: Analyzing research orientation among 3,950 French mathematicians, this paper finds that the gender gap is one of volume rather than quality: women publish 17% fewer papers but achieve comparable citations and journal quality. Women occupy a systematically different position in the mathematical landscape, with greater concentration in higher-output applied subfields masking a within-subfield gap that rises to 25% at comparable positioning.
Agendas of Necessity? Gender, Specialization, and Interdisciplinarity in Mathematics, New project. Funding application in process.
Undisclosed Interests: Prevalence, Determinants, and Consequences of Conflict-of-Interest Disclosure in Biomedical Research, with Nicolas Carayol Funding: ANR ResTores grant (ANR-23-CE36-0006). PI: Dr. Florian Naudet
Summary: In this paper, we leverage a rich dataset combining public disclosures of companies with financial ties to academics, administrative data, firm data, and bibliometric data to assess the prevalence and determinants of conflict of interest disclosure, using state-of-the-art machine learning and AI techniques.