OIT - Revisiting occupational segregation and the valuation of women’s work

X Revisiting occupational segregation and the valuation of women’s work Authors / Hannah Liepmann, Ariane Hegewisch

December / 2025 ILO Working Paper 158© International Labour Organization 2025 Attribution 4.0 International (CC BY 4.0) This work is licensed under the Creative Commons Attribution 4.0 International. See: https:// creativecommons.org/licenses/by/4.0/. The user is allowed to reuse, share (copy and redistribute), adapt (remix, transform and build upon the original work) as detailed in the licence. The user must clearly credit the ILO as the source of the material and indicate if changes were made to the original content. Use of the emblem, name and logo of the ILO is not permitted in connection with translations, adaptations or other derivative works. Attribution – The user must indicate if changes were made and must cite the work as follows: Liepmann, H., Hegewisch, A. Revisiting occupational segregation and the valuation of women’s work. ILO Working Paper 158. Geneva: International Labour Office, 2025.© ILO. Translations – In case of a translation of this work, the following disclaimer must be added along with the attribution: This is a translation of a copyrighted work of the International Labour Organization (ILO). This translation has not been prepared, reviewed or endorsed by the ILO and should not be considered an official ILO translation. The ILO disclaims all responsibility for its content and accuracy. Responsibility rests solely with the author(s) of the translation. Adaptations – In case of an adaptation of this work, the following disclaimer must be added along with the attribution: This is an adaptation of a copyrighted work of the International Labour Organization (ILO). This adaptation has not been prepared, reviewed or endorsed by the ILO and should

not be considered an official ILO adaptation. The ILO disclaims all responsibility for its content and accuracy. Responsibility rests solely with the author(s) of the adaptation. Third-party materials – This Creative Commons licence does not apply to non-ILO copyright materials included in this publication. If the material is attributed to a third party, the user of such material is solely responsible for clearing the rights with the rights holder and for any claims of infringement. Any dispute arising under this licence that cannot be settled amicably shall be referred to arbitration in accordance with the Arbitration Rules of the United Nations Commission on International Trade Law (UNCITRAL). The parties shall be bound by any arbitration award rendered as a result of such arbitration as the final adjudication of such a dispute. For details on rights and licensing, contact: rights@ilo.org. For details on ILO publications and digital products, visit: www.ilo.org/publns.

ISBN 9789220429976 (print), ISBN 9789220429983 (web PDF), ISBN 9789220429990 (epub), ISBN 9789220430002 (html). ISSN 2708-3438 (print), ISSN 2708-3446 (digital) https://doi.org/10.54394/YGCL5095

The designations employed in ILO publications, which are in conformity with United Nations practice, and the presentation of material therein do not imply the expression of any opinion whatsoever on the part of the ILO concerning the legal status of any country, area or territoryor of its authorities, or concerning the delimitation of its frontiers or boundaries. See: www.ilo. org/disclaimer. The opinions and views expressed in this publication are those of the author(s) and do not necessarily reflect the opinions, views or policies of the ILO.

Reference to names of firms and commercial products and processes does not imply their endorsement by the ILO, and any failure to mention a particular firm, commercial product or process is not a sign of disapproval. Information on ILO publications and digital products can be found at: www.ilo.org/researchand-publications ILO Working Papers summarize the results of ILO research in progress, and seek to stimulate discussion of a range of issues related to the world of work. Comments on this ILO Working Paper are welcome and can be sent to liepmann@ilo.org.

Authorization for publication: Caroline Fredrickson, Director, Research Department ILO Working Papers can be found at: www.ilo.org/research-and-publications/working-papers Suggested citation: Liepmann, H., Hegewisch, A. 2025. Revisiting occupational segregation and the valuation of women’s work, ILO Working Paper 158 (Geneva, ILO). https://doi.org/10.54394/YGCL509501 ILO Working Paper 158

Abstract While population ageing increases the demand for care work, new automation technologies, including AI, reinforce the importance of human interaction, with recent research documenting significant wage premiums for social skills. Against this background, we investigate two factors behind the gender wage gap: occupational gender segregation and lower pay in female-dominated occupations, especially care work, where social skills are central. Using CPS data from 1972 to 2024, we show that occupational gender segregation continues to be pronounced in the United States. Many care occupations, in particular, have only seen small increases in male employment shares. Such greater occupational shares of women continue to correlate with significantly lower wages: Conditional on observable characteristics, a 1 percentage point increase in the occupational share of women during 2015-24 was on average associated with a 0.22 percent decrease in wages for women and a 0.20 percent decrease for men. We then analyze whether high returns to social skills are distorted in the care sector, where we hypothesize that workers capture fewer returns on their performance because of the public-goods aspect of care work.

decrease in wages for women and a 0.20 percent decrease for men. We then analyze whether high returns to social skills are distorted in the care sector, where we hypothesize that workers capture fewer returns on their performance because of the public-goods aspect of care work. Based on combined CPS and ONet data, we investigate occupation-level skills returns for 201524. We find that they are indeed sector-dependent: they are insignificant for care workers but sizeable for business services workers. About the authors Hannah Liepmann is an Economist in the Research Department of the International Labour Organization in Geneva and an IZA Research Fellow. Ariane Hegewisch is a Senior Research Fellow at the Institute for Women’s Policy Research in Washington, DC.02 ILO Working Paper 158 Abstract 01 About the authors 01 X Introduction 05 X 1 Data, Main Variables, and Sample Definition 09 X 2 Trends in Occupational Gender Segregation 10 X 3 Female Employment Shares and Occupational Wages 14 X 4 The Valuation of Social Skills 18 4.1 Theoretical Considerations and Hypothesis 18 4.2 Econometric Approach for Estimating Returns to Skills 20 4.3 Findings 22 X Conclusion 28

Appendix: Additional tables and figures (intended for online publication) 29

References 43 Acknowledgements 47 Table of contents03 ILO Working Paper 158 List of Figures Figure 1. The occupational index of gender dissimilarity for all employed individuals and by educational attainment, 1972-2024 11 Figure 2. Change in women’s share of selected occupations, 1972 to 2024 12 Figure 3: Relationship between wages and the occupational share of women, all workers and separately by educational attainment, for different time intervals 17 Figure A.1: The index of dissimilarity by race and ethnicity, 1972-2024 29 Figure A.2: The index of dissimilarity capturing race and ethnic segregation among men and

and separately by educational attainment, for different time intervals 17 Figure A.1: The index of dissimilarity by race and ethnicity, 1972-2024 29 Figure A.2: The index of dissimilarity capturing race and ethnic segregation among men and women, 1972-2024 3004 ILO Working Paper 158 List of Tables Table 1: The impact of the composition and mix effects on the Index of Dissimilarity, 19722024, selected years 13 Table 2: Regression of logged wages on occupational shares of women, 2015-2024 15 Table 3: Regression of logged wages on occupational shares of women, 2015-2024, by gender and educational attainment 16 Table 4: Summary statistics for general characteristics and skills variables, all and by sector, 2015-24 23 Table 5: Regression results capturing the relationship between social skills and wages by sector, all workers during 2015-2024 24 Table 6: Regression results capturing the relationship between social skills and wages by sector, by gender and education, 2015-2024 25 Table 7: Regression results capturing the relationships of social, cognitive and manual skills with wages by sector, all workers during 2015-2024 27 Table A.1: Summary statistics for the sample used 31 Table A.2: The occupational index of gender dissimilarity 1972-2024, for all employed individuals and separately for workers with and without a four-year college degree (values) 32 Table A.3: The index of dissimilarity in comparison to Blau et al. (2013) 34 Table A.4: Definition of care services and business services, following Folbre et al. (2022) 35 Table A.5: Summary statistics for general characteristics and skills variables, by occupational groups, years 2015-2024 37

Table A.4: Definition of care services and business services, following Folbre et al. (2022) 35 Table A.5: Summary statistics for general characteristics and skills variables, by occupational groups, years 2015-2024 37 Table A.6: Regression results capturing the relationship between social skills and wages by broad occupational group, all workers during 2015-2024 38 Table A.7: Regression results capturing the relationship between social skills and wages by sector, for different age groups and by race and ethnicity, during years 2015-2024 39 Table A.8: Regression results capturing the relationship between different social skills and wages by sector, 2015-2024 40 Table A.9: Regression results capturing the relationship between assisting and caring skills and wages by sector, by education during 2015-2024 41 Table A.10: Regression results capturing the relationships of social, cognitive and manual skills – and related interactions - with wages by sector, all workers during 2015-2024 4205 ILO Working Paper 158 X Introduction The U.S. labor market has been characterized by high levels of occupational gender segregation. After occupations had become more gender integrated for at least three decades, further progress stagnated or at least slowed from the mid-1990s onward (Blau, Brummund, and Liu 2013; Blau 2025; England, Levine, and Mishel 2020, Hegewisch et al. 2010). Occupational gender segregation matters economically because wages for both women and men have traditionally been lower in occupations with greater shares of women (Blau and Kahn 2017; Forster et al. 2020; Hegewisch and Hartmann 2014; Olivetti et al. 2024). Female-dominated care work is a pertinent example (e.g., Folbre, Gautham, and Smith 2023). Such undervaluation of female-dominated

Hegewisch and Hartmann 2014; Olivetti et al. 2024). Female-dominated care work is a pertinent example (e.g., Folbre, Gautham, and Smith 2023). Such undervaluation of female-dominated work is consistent with the view that, while individuals’ skills are key for their labor market success, the value that markets and societies attach to skills also depends on who possesses those skills and what is being produced (ILO 2023; Osterman et al. 2022). In the coming decades, population aging is likely to further increase the demand for female-dominated care work at the same time as the demand for male-dominated work in the area of information technology is expected to rise (see ILO forthcoming). There are two alternative scenarios for the impact of these trends on gender equality in the labor market. The first focuses on the relative stability of occupational segregation and the undervaluation of skills associated with women’s work, suggesting that it is unlikely that the labor market will become more equal for women and men. A second scenario, however, is suggested by research highlighting the increasing returns to social skills and human interaction (Deming 2017, among others) and a competitive advantage for women because women are more likely to employ social skills (Cortes, Jaimovich, and Siu 2024; Cortés et al. 2024). In the era of AI, social skills will gain even greater recognition, to the extent that the automation of certain cognitive tasks shifts workers’ time use to human interaction, which cannot be automated (Lee et al. 2025; Shao et al. 2025). Taken together, these latter analyses suggest that gender inequality in wages could decline further as a result of market forces, whereas the first scenario would suggest that absent major change, gender inequality will continue if not worsen. Against this background, our analysis proceeds three-fold. First, we examine recent trends in occupational segregation by calculating the index of dissimilarity at the detailed occupational level and highlighting developments in selected occupations for 1972 to 2024. Second, we examine

gender inequality will continue if not worsen. Against this background, our analysis proceeds three-fold. First, we examine recent trends in occupational segregation by calculating the index of dissimilarity at the detailed occupational level and highlighting developments in selected occupations for 1972 to 2024. Second, we examine the economic significance of occupational segregation by relating individuals’ annual wages to the share of women working in their occupation. In these two parts, we build on our earlier work with Jeffrey Hayes and Heidi Hartmann (Hegewisch et al. 2010; see also Hegewisch and Liepmann 2013). Third, we use a skills-based approach and compare the occupation-level remuneration of social skills, as well as cognitive and manual skills, in female-dominated care work and more gender-balanced business services, focussing on years 2015 to 2024. As discussed below, this analysis extends Folbre, Gautham, and Smith (2023) and Pietrykowski (2017). Using harmonized data from the March Annual Social and Economic Supplement of the Current Population Survey (CPS) (Flood et al. 2024), we find that after stagnating from the mid-1990s onward, occupational gender segregation started to decline again from 2010 onward, albeit at a modest pace. There was a more notable further decline after 2020, associated with the COVID-19‒related recession. In 2024, the index of dissimilarity equalled 46.8 (compared with 51.9 and 2010 and 51.8 in 2000), which represents the share of women or men who would have to switch occupations for their share in each occupation to correspond to their share in overall employment. Many care occupations, in particular, continue to employ large shares of women, with only small increases in male employment in more recent years. In contrast, many managerial06 ILO Working Paper 158 and professional occupations – often in the broadly understood area of business services - have witnessed a marked increase in female employment shares over time. At the other side of the

with only small increases in male employment in more recent years. In contrast, many managerial06 ILO Working Paper 158 and professional occupations – often in the broadly understood area of business services - have witnessed a marked increase in female employment shares over time. At the other side of the spectrum, more technical occupations in construction, manufacturing, and IT-related fields remain male-dominated. We subsequently show that the relationship between the occupational share of women and workers’ annual wages continues to hold. Controlling for educational attainment, age, working time, and general wage differences between women and men, among other characteristics, we find that a 1 percentage point increase in the occupational share of women is associated with a 0.21 percent decline in annual wages for the pooled years 2015 to 2024. The same estimated decline is similar for female workers (-0.22 percent) and male workers (-0.20 percent). Over the decades studied, the negative association between wages and the share of women working in an occupation has remained remarkably unchanged. Analyzed for different educational groups, the relationship has become more negative for college-educated workers, while becoming slightly less pronounced for workers without a college degree. To explore occupation-level skills returns as one mechanism underlying lower wages for female-dominated work, we then focus on the returns to skills in the care sector compared to business services. The care sector comprises diverse health, education, and social assistance occupations with varying entry requirements and wage levels. During years 2015 to 2024, 75.3 percent of care workers were female and the sector accounted for a large share of total female employment (41.1 percent). Business services – defined as comprising finance, insurance, real estate, professional, scientific and management industries, and public administration - are instead more gender balanced (46.8 percent were female and the sector accounted for 25.6 percent of

total female employment). Because the care sector and the business services sector have similar educational requirements but different gender compositions, comparing the two yields informative insights into the broader theme of the valuation of women’s work (Folbre et al. 2023). We are particularly interested in the returns to social skills because the labor economics literature stresses their growing importance: In Deming's (2017) model, social skills improve workers’ productivity in team contexts, resulting in higher wages for more socially skilled workers. We hypothesize, however, that this positive effect of social skill use on wages is distorted for care workers because of a number of well-established factors leading to the undervaluation of care work. This includes its characteristic as a public good, which makes it more difficult for care workers to capture the returns to their performance (Dwyer 2013; England 2005a; England et al. 2002; Folbre et al. 2023). To test this hypothesis, we merge the CPS data with ONet data (Carpenter et al. 2022), focusing on pooled years 2015 to 2024. Following Deming (2017), we construct a social skill index based on occupational skill requirements for coordination of action, negotiation, persuasion, and social perceptiveness. We additionally account for cognitive skills (as in Acemoglu and Autor 2011) and manual skills (as in Cortes, Jaimovich, and Siu 2024). We then regress annual wages on these skill variables interacted with different sector indicators and include the same control variables as in the previous wage analysis. Consistent with our hypothesis, we find that a one standard-deviation increase in occupation-level social skill use increases annual wages in care services by 2.3 percent, but the effect is statistically insignificant. The same effect is precisely estimated and around 2.5 times larger (at 5.8 percent) for business services. The pattern is remarkably similar for women and men, workers with different levels of educational attainment, most race and ethnicity groups, and for the individual social skills components of the index. In contrast, cognitive

around 2.5 times larger (at 5.8 percent) for business services. The pattern is remarkably similar for women and men, workers with different levels of educational attainment, most race and ethnicity groups, and for the individual social skills components of the index. In contrast, cognitive skills are associated with relatively large and statistically significant wage returns for both care services and business services. Manual skills are associated with smaller and less statistically07 ILO Working Paper 158 significant wage returns in both sectors. Therefore, social rather than cognitive and manual skills cause the differential occupation-level skills returns between care services and business services. This finding is robust in accounting for skills bundles, specifically the joined occupational requirements of social and cognitive as well as social and manual skills. Our paper relates to the already mentioned studies that investigate trends in occupational gender segregation until 2009 (Blau et al. 2013; Hegewisch et al. 2010) and, at the more aggregated level, until 2017 (England et al. 2020). We replicate the main conclusions of these studies and provide an updated analysis of occupational segregation until 2024, analyzing whether within or between-occupational effects drive the changes, presenting results for different educational levels and race and ethnicity groups, and tracking the relationship between female occupational shares and wages. Our main contribution is to analyze how differential returns to skills help explain the undervaluation of female-dominated work, using the example of the care sector. A rich literature investigates the low remuneration of care work compared to other type of work (e.g., England and Folbre 1999; England, Budig, and Folbre 2002; England 2005; Dwyer 2013; ILO 2018; Osterman et al. 2022; Folbre, Gautham, and Smith 2023). Folbre, Gautham, and Smith (2023) contrast care work with business services and find that the former pays significantly less despite similar levels of educational attainment. We apply and extend these authors’ approach to study skills returns.

et al. 2022; Folbre, Gautham, and Smith 2023). Folbre, Gautham, and Smith (2023) contrast care work with business services and find that the former pays significantly less despite similar levels of educational attainment. We apply and extend these authors’ approach to study skills returns. In this sense, our analysis is inspired by wage setting contexts aimed at achieving comparable worth, i.e., equal pay for work of equal value, that have long recognized the importance of accounting for skills in addition to educational attainment and other job demands (see for example Wage Equity Study Team 2023). Our skills analysis also complements Pietrykowski (2017), who shows that caring and assisting skills in the United States are associated with a wage penalty in lower-wage occupations and a wage premium in higher-wage, male-dominated occupations. We instead investigate the returns to social skills in different sectors and occupations that encompass a range of wage levels. Our skills analysis is likewise motivated by the growing literature that stresses the importance of social skills (Deming 2017; Deming and Kahn 2018; Weinberger 2014). Several studies analyze these developments from a gender perspective, demonstrating that the increased demand for social skills has enhanced women’s labor market prospects. Black and Spitz-Oener (2010) show for Germany that women’s greater use of social (and analytical) skills correlated with a narrowing of the gender wage gap between 1979 and 1999. For the United States, Borghans, Ter Weel, and Weinberg (2014) find that the rising returns to social skills mirror the closing of the gender wage gap up until 1992. Cortes, Jaimovich, and Siu (2024) link the growing reliance on social skills in high-wage occupations to more women entering such occupations between 1980 and 2016. Finally, Cortés et al. (2024) provide causal estimates that women in the United States were better able to adapt to technological change than men, as between 1980 and 2017 women moved into higher-paying occupations that require social (and abstract) tasks and thus improved their relative wages.

Finally, Cortés et al. (2024) provide causal estimates that women in the United States were better able to adapt to technological change than men, as between 1980 and 2017 women moved into higher-paying occupations that require social (and abstract) tasks and thus improved their relative wages. We complement this literature by showing that the occupation-level returns to social skills are not uniform across the labor market. Our findings suggest that the previously mentioned studies capture longer-term shifts towards greater gender integration of initially male-dominated occupations that narrowed the gender wage gap. In contrast, workers in more female-dominated care work do not benefit to the same extent from the occupational returns to social skills. In combination with the slow shift towards greater gender integration of occupations in recent years, this makes a rapid closing of the remaining gender wage gap unlikely.08 ILO Working Paper 158 In the following Section 1, we describe our data, main variables, and sample. Section 2 presents trends in occupational gender segregation, followed by our analysis of the link between female employment shares and occupational wages in Section 3. Section 4 includes our analysis of sectoral wage returns to social skills as well as cognitive and manual skills. Section 5 summarizes and briefly discusses broader implications for the expected evolution of the gender wage gap.09 ILO Working Paper 158 X 1 Data, Main Variables, and Sample Definition

We analyze data from the March Annual Social and Economic Supplement (ASEC) of the CPS, which is representative of the civilian, noninstitutionalized population. It provides cross-sectional, annual information over a long timeframe, covering demographic and personal characteristics and comprehensive employment outcomes, such as an individual’s employment status, occupation, and wages in the previous calendar year. We use the harmonized version of these data as provided by IPUMS (Flood et al. 2024) and focus on years 1972 to 2024. An advantage of the CPS data relates to the availability of a harmonized classification of occupations, which allows us to assess longer-term, annual trends in occupational gender segregaas provided by IPUMS (Flood et al. 2024) and focus on years 1972 to 2024. An advantage of the CPS data relates to the availability of a harmonized classification of occupations, which allows us to assess longer-term, annual trends in occupational gender segregation with the same underlying data source. Specifically, we use the detailed occupational variable that is based on the 2010 Census Bureau’s ACS classification standard and was harmonized across time by Flood et al. (2024). These researchers achieved the most consistent occupational categories across surveys, grouping more detailed occupations together when these were not separately elicited in previous years. They confirmed the robustness of their groupings through a trends analysis of key occupational outcomes (IPUMS 2025). Our analysis considers employed individuals in the civilian labor force, aged 25 to 64. We disregard very small occupational categories with less than 20 observations in a given year, which leads us to exclude 1.4 per cent of the initial sample. The final sample consists of around 3,250,000 observations, employed in on average 294 different occupations per year. We present summary statistics for this sample in Appendix Table A.1. Throughout the analysis, we weight results by ASEC person weights. To classify occupational skills requirements, we merged the CPS data with ONet data. We rely on the 26.1 version of ONet from November 2021 as provided by Carpenter et al. (2022), because it contains a crosswalk between the highly detailed occupational classification of ONet and the 2010 Census Bureau’s ACS occupational classification standard. We employ this crosswalk with minor modifications.1 For our analysis of occupational skills returns, we restrict the attention to years 2015 to 2024, because the 2021 ONet skills data is likely inaccurate for earlier years. We focus on social skills but also create measures of cognitive and manual skills. The details of the variable construction are provided in Section 4. 1 Specifically, we needed to aggregate and cross-walk a few occupations to match the ONet data with the harmonized occupational

focus on social skills but also create measures of cognitive and manual skills. The details of the variable construction are provided in Section 4. 1 Specifically, we needed to aggregate and cross-walk a few occupations to match the ONet data with the harmonized occupational variable of Flood et al. (2024).10 ILO Working Paper 158 X 2 Trends in Occupational Gender Segregation

In this section, we show that occupational gender segregation remains widespread, using Duncan and Duncan's (1955) index of dissimilarity as a summary measure for the overall extent of occupational gender segregation in the United States. The index is defined as follows: D 0
5 J j 1





M j M 





W j W (1) where M j and W j are the number of men and women working in a detailed occupation j, respectively, and M and W reflect the male and female population size in the labor market. The index takes the value of zero when the share of women (and men) in each detailed occupation equals the share of women (and men) in total employment (complete gender integration). It takes the value of one when occupations are entirely segregated by gender. Intermediate values indicate the share of women (or men), who would need to change occupations, for all occupations to be fully gender balanced. In 2024, the value of the index for the civilian labor force aged 25 to 64 was 46.8, indicating that almost half of all employed women and men would have to change occupations for the gender distribution in each occupation to equal that in the overall workforce. While the 2024 value of the index is the lowest measured since 1972, the trends in the index over the last five decades show a substantial change in the pace of occupational gender integration (see Figure 1 and Appendix Table A.2 with the underlying values). In the 1970s to 1990s, the value of the index for

the index is the lowest measured since 1972, the trends in the index over the last five decades show a substantial change in