Large-scale career data show that women artists face unequal exhibition opportunities and auction access, and that an artist's institutional environment predicts market success more strongly than the artist's own gender — but all of this is descriptive association. The open problem posed by Wang, Gates, Resch and Barabási's study of 65,768 contemporary artists and 20,389 institutions is to move from this quantified description to causal explanation: identify the specific decision-making policies, curator behaviors, and collector practices that generate the measured institutional gender inequalities, and quantify how the historical gender imbalance crystallized in legacy museum collections propagates into imbalanced exposure for visitors.

Background

The quantitative study of the art world treats exhibitions, institutional holdings, and auction sales as career data. In this framework an artist's career is a sequence of exhibitions at institutions (galleries and museums), and the secondary market is accessed through auctions. Wang, Gates, Resch and Barabási (2025 preprint; published 2026 in Nature Communications) analyzed a commercial dataset (Artfacts.net, with pricing from magnus.net) covering the exhibition careers of 496,354 artists at 23,570 institutions, restricted to 65,768 contemporary artists who began exhibiting in or after 1990, at 20,389 institutions worldwide. Gender was assigned as a binary label using genderize.io, a name-based gender-inference service, supplemented by expert-curated labels; such services have documented, ethnicity-dependent coverage and accuracy limits, as benchmarked by Santamaría and Mihaljević (2018). The study distinguishes two criteria of gender equity: gender-neutrality, meaning that artists' access to exhibition opportunities at an institution is statistically independent of artist gender, and gender-balance, meaning approximate gender parity in the artists actually shown. Each institution was classified by comparing observed exhibition counts against the artist-population gender ratio using Bayes-factor hypothesis testing (a Bayes factor is the likelihood ratio between two hypotheses; institutions in the anecdotal range 1/3 < BF < 3 were left unclassified). The headline results: 58% of institutions are gender-neutral but only 24% gender-balanced; man-overrepresentation increases with institutional prestige; and the auction market shows the sharpest disparity. To connect institutions to careers, the authors define an artist's co-exhibition gender, a network-derived measure of the gender composition of the institutions where an artist exhibits. A logistic regression predicting access to the auction market finds that co-exhibition gender correlates with market access more strongly than the artist's own gender, and co-exhibition gender is stable over careers, suggesting lock-in. This complements two adjacent strands: collection-demographics studies, notably Topaz et al. (2019), which measured the gender and ethnic composition of permanent holdings of major U.S. museums and found them heavily dominated by men, especially in historical (legacy) collections; and art-market economics, including hedonic price studies (regressions of prices on artwork and artist attributes) by LeBlanc and Sheppard (2021) on gender, ethnicity and origin in contemporary prices, and Adams, Kräussl, Navone and Verwijmeren (2021) on gendered auction prices, which debate whether observed gaps reflect market discrimination or selection. What none of these provides is causal, decision-level explanation. Wang et al. explicitly state that their methods are not designed to identify the specific policies and decisions that create and perpetuate the barriers, note that the source of institutional inequality cannot be fully understood without data on curators and collectors — the actors who drive acquisition, exhibition, and auction outcomes — which does not yet exist at scale, and flag the mirror-image historical question: the gender inequality crystallized in legacy museum collections and its consequential imbalanced exposure for visitors, which their contemporary-career focus cannot address.

Problem Statement

Identify, with empirical causal evidence, the specific decision-making policies, curator behaviors, and collector practices that generate the institutional gender inequalities quantified by Wang et al. — the man-overrepresentation that grows with institutional prestige, the gender-dependence of exhibition access, and the co-exhibition-gender lock-in that correlates with auction-market access — and quantify how the historical gender imbalance crystallized in legacy museum collections propagates into imbalanced exposure for museum visitors. Meeting this target means moving beyond descriptive association: connecting measured outcomes to identified decision-level mechanisms and quantifying the exposure consequences of legacy-collection composition, robust to the documented limitations of binary, name-based gender annotation and incomplete institutional coverage.

References

  1. Xindi Wang, Alexander J. Gates, Magnus Resch, Albert-László Barabási, 'Quantifying institutional gender inequality in contemporary visual art', Nature Communications 17:6026 (2026), DOI 10.1038/s41467-026-71470-4, https://doi.org/10.1038/s41467-026-71470-4 (preprint arXiv:2506.22103, 2025, DOI 10.48550/arXiv.2506.22103, https://arxiv.org/abs/2506.22103)
  2. Chad M. Topaz, Bernhard Klingenberg, Daniel Turek, Brianna Heggeseth, Pamela E. Harris, Julie C. Blackwood, C. Ondine Chavoya, Steven Nelson, Kevin M. Murphy, 'Diversity of artists in major U.S. museums', PLOS ONE 14(3):e0212852 (2019), DOI 10.1371/journal.pone.0212852, https://doi.org/10.1371/journal.pone.0212852
  3. Abigail LeBlanc, Stephen Sheppard, 'Women artists: gender, ethnicity, origin and contemporary prices', Journal of Cultural Economics 46(3):439–481 (2022), DOI 10.1007/s10824-021-09431-6, https://doi.org/10.1007/s10824-021-09431-6
  4. Renée B. Adams, Roman Kräussl, Marco Navone, Patrick Verwijmeren, 'Gendered Prices', The Review of Financial Studies 34(8):3789–3839 (2021), DOI 10.1093/rfs/hhab046, https://doi.org/10.1093/rfs/hhab046
  5. Lucía Santamaría, Helena Mihaljević, 'Comparison and benchmark of name-to-gender inference services', PeerJ Computer Science 4:e156 (2018), DOI 10.7717/peerj-cs.156, https://doi.org/10.7717/peerj-cs.156
  6. Nanne van Noord, 'Operationalising the subjective: Coding visual art with AI', Current Opinion in Psychology 72:102386 (2026), DOI 10.1016/j.copsyc.2026.102386, https://doi.org/10.1016/j.copsyc.2026.102386