Abstract
Drawing on data from the Australian Bureau of Statistics and the Household, Income and Labour Dynamics in Australia Survey, this paper assesses whether women's position in the Australian labour market has improved since the early 2000s. Women have increased their educational attainment and participation in paid work, yet progress toward gender equality remains limited. Women continue to be less likely than men to work full-time, more likely to experience underemployment and financial stress, and remain concentrated in low-paid, insecure and highly feminised sectors. Occupational and industrial segregation persists, and the gender pay gap has narrowed only modestly despite rising qualifications, reflecting entrenched patterns of gender undervaluation and gender bias in wage setting arrangements. Although women's educational gains have contributed to narrowing of the gender wage gap, declining returns to education and experience has limited progress. These findings highlight the continuing influence of structural and institutional forces on gendered labour-market outcomes and echo longstanding critiques of market-based approaches to gender equality. They also highlight the critical role of gender undervaluation assessments.
Introduction
In her influential article ‘All Change, Still Gendered: The Australian Labour Market in the 1990s’ (1998), Barbara Pocock urged policymakers and employment relations scholars to take seriously the issue of time, including the time required for unpaid care work, the degree of worker control over start and finishing times in paid work, and the consequences of working less than full-time hours. Writing in the 1990s, Pocock was especially critical of the market-based solutions to gender inequality that dominated at the time, including labour market deregulation, individual bargaining and flexible work arrangements such as casual employment. Her contention was that these approaches are gender-blind and ignore ‘…the interdependent economies of time/work – across the home, community and the paid workforce…’ (p.584). She warned that unless the interdependence between paid and unpaid work was acknowledged, women's labour market disadvantage would become further entrenched.
In support of her critique, Pocock examined women's labour-market experiences during the 1990s, focusing on indicators shaped by the interdependence of home and waged work, including employment patterns, occupational segregation, wages and the quality of working life. Across these indicators, she found little evidence that women's outcomes had improved relative to men's over during the period studied. Her central concern was the growing dependence of women on part-time casual work as a strategy for managing paid work and unpaid responsibilities. Pocock argued that such forms of ‘marginal’ employment left women increasingly exposed to economic insecurity and long-term financial risk. She was especially critical of the industrial relations system that had enabled the expansion of casual work. Although casual loadings were intended to deter such employment forms, they had not prevented the rapid expansion of this employment arrangement.
In the period since Pocock's article was published, Australia has undergone significant social and economic change, including a marked increase in women's labour-force participation and substantial growth in the proportion of Australians holding tertiary qualifications. In 2000, for example, 16.2% of women and 15.2% of men aged 15–64 held a tertiary qualification. By 2024, these shares had increased to 38.8% and 30.7%, respectively (see Figure S1 in the supplementary online appendix). In 2005 the female labour force participation rate was equal to 68.1% and the gender participation gap was equal to 14.9 percentage points. By 2025, the former had climbed to 77.7% and the gender participation gap had narrowed by 8.7 points to 6.2 percentage points (Table 1).
Labour force characteristics, women and men, various years.
Note: The Labour Force Participation (LFP) rate measures the number of people employed plus unemployed as a share of the civilian population. The Employment/Population (E/P) rate measures the number of people employed as a share of the civilian population. A person is considered full-time if they work 35 h or more per week across all jobs. Symbol * indicates number should be treated with caution.
Source: ABS Labour Force, Detailed (Cat. No. 6291.0.55.001, Table 1). Reference period April 2025. Released 22 May 2025. Original Series.
These participation trends are not unique to Australia (OECD 2025, 2025a; Rubery, 2024). Between 2005 and 2025 in the Euro area (20 countries), the gender participation gap narrowed by 7.7 percentage points. Similar or larger reductions were observed in New Zealand (7.9 points), the Netherlands (8.3 points), Ireland (12.4 points), and Spain (14.8 points) (OECD 2025b). Globally, however, the burden of unpaid work continues to fall disproportionately on women. Women, on average, perform almost twice as much unpaid work as men (OECD 2025a, p.112). In Australia, the gender gap in unpaid work is large at around 1 h and 19 min per day (ABS 2022). 1 In the Nordic countries the gaps are smaller (49 min in Sweden; 57 min in Denmark and 59 min in Norway) (OECD 2025c). The difference reflects, in part, care giver models, social norms and welfare regimes. Women in Nordic countries tend to be more fully integrated into paid employment, supported by leave entitlements and well-funded welfare regimes (Hebson and Rubery, 2018).
Drawing on data from the Australian Bureau of Statistics (ABS) and the Household, Income and Labour Dynamics in Australia (HILDA) Survey, this paper examines whether women's labour market position has improved since the early 2000s, particularly given their significant education investments. In so doing it provides an empirical update to the issues highlighted by Pocock in the late 1990s. While the analysis is primarily descriptive, it is situated within an institutional understanding of how gender norms, care responsibilities and labour market structures interact to affect women's labour market outcomes. The aim of the paper is to provide, not only a critical evidence base on women's labour market experiences and the persistent structural barriers they face, but also a valuable benchmark for evaluating progress under the Working for Women strategy (Australian Government, 2024) and other recent policy reforms aimed at enhancing gender equality (e.g., the Fair Work Legislation Amendment (Secure Jobs, Better Pay) Act 2022 [the SJBP Act]).
The remainder of the paper is organised as follows. It begins with a summary of the theoretical framing that underpins the empirical analysis, along with an overview of recent policy initiatives aimed at improving women's labour market outcomes. The empirical analysis then examines work patterns, segregation, wages and quality of working life, following Pocock (1998).The paper concludes with a summary and policy discussion.
It is important to note that an intersectional analysis is not offered. Beyond issues of scope, this decision reflects the conceptual challenges involved in integrating intersectionality into industrial relations research. As McBride and Rodriguez (2024) observe, the relationship between intersectionality and the IR field remains ‘ambiguous, undefined and full of tensions’ (p. 137). Although intersectionality is beginning to appear in IR scholarship, they argue that it risks becoming a ‘stand-alone concept’ detached from its origins in critical race theory, thereby depoliticising the approach (p. 138).
Theory framing and recent policy developments
A substantial body of feminist economic and employment-relations scholarship underscores that labour markets are ‘gendered institutions’ shaped by norms, practices and systems of social reproduction (Elson, 1999). As Pocock (1998) highlights, labour markets operate at the intersection of paid work and unpaid work; they are structured by expectations about who undertakes care, how work is valued, and which jobs are considered compatible with caring. Himmelweit (2007) likewise stresses that decisions about employment and caring are intertwined ‘… so that no theory of the labour market, nor any labour market policy, can realistically ignore caring’ (2007, pp. 582–583).
Hebson and Rubery (2018) provide an excellent review of the contributions of feminist scholarship to employment relations. Among other insights, they show how the concentration of women in particular employment forms such as part-time work can place downward pressure on the employment standards of full-time workers, for example by weakening entitlements to overtime or evening-work penalties. They argue that these effects are less pronounced where women are more fully integrated into waged work and where collective bargaining arrangements limit the scope for undercutting employment standards (p. 151). They also remind us that ‘one of the main causes of the gender pay gap is the tendency to undervalue women's work’ and that ‘much of the undervaluation of women's work is associated with gender segregation…’ (p. 154).
In their discussion of working-time arrangements, Hebson and Rubery argue that the traditional model of ‘standard’ employment (continuous, full-time work) has long acted to exclude many women. However, they also warn that the expansion of diverse, non-standard working-time options (e.g., casual employment) brings new risks, especially for women. They note that in some countries women are increasingly preferencing full-time roles over part-time roles, partly because full-time work typically offers more predictable hours, while part-time jobs are often associated with irregular scheduling and poorer working conditions.
Pocock's (1998) analysis shows that the needs of caregivers were largely overlooked in the labour-market reforms of the 1980s and 1990s. Although Pocock and other feminist scholars argued for labour market policy to afford more attention to the interdependence of home and waged work, the decades since have seen only limited progress. One major exception was the introduction of a national Paid Parental Leave (PPL) scheme in 2011. Prior to the Paid Parental Leave Act 2010, access to paid maternity leave was uneven and largely confined to full-time employees (Productivity Commission, 2009).
A second important policy development was the creation of a legal right for eligible employees to request flexible working arrangements. This right, available to permanent employees and regular, ongoing casuals with at least 12 months of service who met specific criteria such as caring for a school-aged child, was established under the National Employment Standards (NES) of the Fair Work Act 2009 and took effect on 1 January 2010. The 2009 Fair Work Act also introduced the ‘low-paid bargaining stream’, designed to facilitate agreement making and lift wages for low-paid workers, many of whom are women. This stream, however, ‘…was a complete failure as not a single agreement was achieved’ (Charlesworth and Macdonald 2023, p.406).
After a prolonged period of relative inaction on gender equality, the federal Labor Government elected in May 2022 adopted a more proactive stance in this area. A key development was the passage of the Fair Work Legislation Amendment (Secure Jobs, Better Pay) Act (the SJBP Act) in December 2022. The 2022 SJBP Act aims, among other things, to address structural drivers of inequality, including the gender pay gap. Notably, the Fair Work Commission (FWC) is now explicitly required to consider the promotion of gender equality and job security when exercising its functions. Significantly, when making decisions about the setting of minimum award wages, the FWC is required consider the need to achieve gender equality in the workplace by ensuring equal remuneration for work of equal or comparable value, eliminating gender-based undervaluation of work and providing workplace conditions that facilitate women's full economic participation. The FWC has similar obligations in respect to its consideration of annual adjustments to the National Minimum Wage (Charlesworth and Macdonald, 2023; Ellem et al., 2025).
In addition to the above, the FWC may now initiate equal remuneration and work value hearings and make equal remuneration orders (ERO) and findings of gender-based undervaluation without first requiring an application from any of the parties. The amendments also make it clear that, when conducting assessments of equal remuneration and work value, the FWC is not required to do so with reference to a historically male dominated occupation or industry. This particular amendment was in response to concerns that the FWC had applied a conservative interpretation of the FW Act in its decision on a union equal remuneration claim in early childhood education and care (FWC 2015). One of the chief concerns was the FWC required that, for an ERO to be made, the application had to identify a group of male employees doing comparable work (Smith and Whitehouse, 2020).
In addition to the 2022 SJBP amendments, the government has strengthened pay transparency through two main measures: provisions in the SJBP Act prohibiting pay secrecy clauses, and new reporting requirements mandating that the Workplace Gender Equality Agency publish employer-level gender pay gaps for organisations with 100 or more employees from February 2024. The government has also taken steps to reduce barriers to women's workforce participation by expanding childcare subsidies from July 2023. These initiatives are part of a broader gender equality agenda outlined in Working for Women strategy, released in 2024 (Australian Government, 2024). The latter sets the ambitious and laudable goal of achieving gender equality by 2034.
Work patterns
Table 1 presents data on the labour market outcomes of men and women. Comparisons are made at three intervals: April 2005, April 2015 and April 2025. April was selected as the reference point because at the time of writing it represented the most recent labour force data available. The years 2015 and 2005 were chosen to allow for comparisons at 10-year intervals. The data show that, while there was little change in the work patterns of women and men between 2005 and 2015, the pace of change has accelerated since 2015 decade. For example, among women the employment/participation (E/P) rate (which measures the number of persons employed as a proportion of the civilian population rose by 2.2 percentage points to 66.6% between 2005 and 2015. Between 2015 and 2020 it increased by 7.9 points to 77.7%. Similarly, the share of employed women working full-time (35 + hours per week in all jobs) declined slightly in the first period (–0.6 points to 54.1%) but rose by 3.7 points (to 57.8%) between 2015 and 2025. Disaggregated E/P rates by age show the most substantial growth in employment participation occurred among women aged 35–44, with their E/P rate increasing from 71.2% in 2015 to 81.7% in 2025. The recent growth in full-time employment is particularly noteworthy given the long-standing stability in women's full-time employment shares since 2000 and even earlier (discussed further below).
Full-time work
In 2002, in a public lecture titled ‘Can this be the Promised Land? Work and Welfare for the Modern Woman’, Professor Bob Gregory, an eminent labour economist, reflected on why women's rising educational attainment had not translated into higher full-time employment rates. From a neoclassical perspective, higher education should strengthen women's labour market attachment. He argued that ‘The failure of full-time employment to increase over the last 35 years [1965–2000] suggests that investing in more education in anticipation of being employed for more full-time years was either not the intention of young women or, as yet, higher levels of education have not been a good investment’ (Gregory, 2002, p.45).
In his lecture, Gregory examined whether compositional changes might explain the apparent stagnation in women's full-time employment. He considered a range of possibilities, including whether some women were extending their full-time careers while others shifted into part-time work, whether more educated women were displacing their less-educated peers, or whether single women were increasing their share of full-time jobs. His empirical work showed that, across the 1980s and 1990s, the main source of growth in full-time employment was among married women. At an aggregate level the rate of full-time employment among all women remained largely unchanged. These findings led him to conclude that ‘Education is not playing an aggregate role of any significance in adding to the stock of full-time jobs for women or changing the allocation of which women are employed full-time’ (Gregory, 2002, p.46). He attributed this stagnation to structural and policy constraints, including welfare and income-support settings that channelled women into part-time work. In other words, and consistent with Pocock (1998), Gregory's analysis highlighted the institutional and normative factors shaping women's employment patterns during this period.
In the analysis below labour force data from the ABS and HILDA are used to shed light on patterns of full-time work since 2000. If institutional and normative arrangements that shape women's employment patterns have changed, this should be reflected in a change in the way they participate in paid employment.
One challenge when examining full-time employment using ABS labour force data is that a person is considered full-time if they work 35 or more hours per week across all jobs. This means that a person who works 35 or more hours over two part-time jobs will be classified as full-time. This matters as recent years have seen a rise in the incidence of multiple jobholding (MJH). Moreover, the MJH rate is higher among women than men (7.6% vis 6%) and is particularly high among persons who are community and personal service workers in their main job (9.9%) (ABS, 2025). The gender gap in MJH is primarily driven by women's greater levels of underemployment in the primary job (Preston and Wright, 2020).
To overcome the definitional issue, Figure 1 focuses on average hours worked in the main job only. The estimates show that between September 2001 and March 2025, total employment among women and men rose by 2.4 million and 2.1 million, respectively. In percentage terms, women's employment grew by 63.8%, and men's by 45.1% (the smaller growth for men reflecting their higher starting base). Over the same period there was no change in average (main job) hours worked among women. In September 2001 employed women, on average, worked 30.3 h in their main job. By March 2025 this figure stood at 30.4 h. Men, on the other hand, have seen a significant decline in their average working hours, from 41.2 in September 2001 to 37.0 h by March 2025. Consistent with Gregory's analysis of data from the 1980s and 1990s, these data suggest no obvious shift towards full-time employment.

Total employment and average hours worked (main job) by sex, September 2001 to March 2025.
Following Gregory (2002), Figure 2 presents data on the E/P ratios of women disaggregated by age, marital status and hours worked (full-time/part-time status). Focusing first on trends in the part-time (PT) E/P shares (bottom row of Figure 2), the data show very little change in the part-time E/P shares between 2001 and 2019. The exception is among not-married women aged 15–24 where the part-time E/P share has increased sharply in the last 10 years. This likely reflects a greater share of this group combining part-time work and study.

Employment to population ratios of women by age, marital status and full-time/part-time employment status.
The main change is in the full-time (FT) E/P shares of married women, especially those aged 35–54. Between 2001 and 2019 their FT E/P shares were relatively constant and since 2019 they have increased by 7 percentage points to 36% among those aged 35–44 and by 5 percentage points to 37% among those aged 45–54.
While the disaggregated analysis reveals some important trends, what remains striking is the relative stability in full-time E/P shares between 2001 and 2019, despite significant growth in women's educational attainment during this time. Consistent with Gregory's findings for the 1980s and 1990s, there is little evidence that rising education levels since 2000 led to any structural changes in women's employment patterns over the first two decades of this century. This may reflect the highly gendered social and cultural norms that shape education and training choices, career trajectories and pathways from school onwards (Skills Australia, 2025). These norms continue to reproduce stereotypes about the kinds of jobs appropriate for women and men and associated working arrangements.
The recent rise in women's full-time employment since 2019 is unlikely to reflect long-term shifts in educational investment, as women's educational attainment has been high and increasing for decades without producing a corresponding increase in full-time work. Explaining the change is beyond the scope of this paper, but several plausible demand and supply side factors may be at play. On the demand side it may be that skill shortages and historically low unemployment have increased employers’ need for labour and more women are working longer hours in their main job (35 or more per week). On the supply side, cost-of-living pressures and household budget constraints may be encouraging more married women to work full-time. Another possibility is that women are increasingly opting for full-time rather than part-time work because remote and hybrid work have expanded opportunities to combine paid and unpaid responsibilities (Birch and Preston, 2025). It may also be that full-time work has become more attractive due to greater predictability and stability in working hours (Hebson and Rubery, 2018).
To further unpack the patterns in Figure 2, a multinomial logit regression using HILDA data is estimated. The dependent variable is a three-category employment status indicator: 0 for not employed (base category), 1 for employed full-time in the main job, and 2 for employed part-time in the main job. The sample is restricted to individuals aged 25–64 to better capture trends among those likely to have completed tertiary education. The decision to use a multinomial logit rather than a two-stage selection model is that the former supports an understanding of the determinants of labour market status, that is, what factors predict whether someone is employed full-time, employed part-time or not employed.
Three sets of marginal effects are reported, one for men employed full-time, one for women employed full-time and one for women employed part-time. In the interest of space, these estimates are located in Table S1 in the supplementary online appendix. The regression controls for age (three dummy variables, with ages 25–34 as the reference group), education (a binary variable equal to 1 if the respondent has a Bachelor degree or higher), marital status (five dummies covering married, de facto, separated, divorced and widowed; the reference group is never married) and non-labour income. The regression also includes controls for presence and number of dependent children, whether a household member has a disability, geographic factors (urban/rural and state) and four time-period dummies covering 2006–2010, 2011–2015, 2016–2019 and 2020–2023 (with 2001–2005 as the reference period). To assess whether the effects of tertiary education, marital status and non-labour income on full-time employment have changed over time, interactions between each of these variables and the period dummies are included in the specification.
Holding other factors constant, the results for men indicate that the likelihood of being employed full-time (relative to not being employed) is positively associated with having a degree, being married and having dependent children. The coefficients on the period dummies show a decline in men's probability of full-time employment over 2020–2023. The period interactions further indicate that, from 2011 onwards, marriage has become increasingly associated with full-time employment for men. No causal inference is implied. The effects of education and non-labour income on the probability of male full-time employment have remained stable over the past 23 years.
For women, degree holders are 14.4 percentage points more likely than non-degree holders to be in full-time employment rather than not employed (the base category), holding other factors constant. Married women, by contrast, are 5.4 percentage points less likely than never-married women to be employed full-time. Non-labour income also matters. For every $1000 increase in gross household weekly income (excluding her own earnings), the probability of full-time employment falls by 1.52 percentage points. The interaction terms show no change over time in the effect of educational attainment on the likelihood of women working full-time. This mirrors Gregory's (2002) earlier findings. A detailed analysis of why tertiary education has failed to generate corresponding employment gains lies beyond the scope of this paper. However, as noted, from a feminist institutional perspective it is consistent with strong normative forces affecting the employment decisions and participation patterns of women (Pocock, 1998).
Consistent with the earlier descriptive analysis, the 2020–2023 marriage interaction terms in the female regressions show, as with men, that marriage has become increasingly associated with full-time employment. These developments effectively reverse the earlier negative relationship between marriage and women's full-time work. The non-labour income interactions for the same period also indicate that the negative effect of non-labour income on full-time employment has weakened. During 2020–2023, the marginal effect is −0.008 (–0.015 + 0.007), implying that the probability of full-time employment falls by only 0.8 percentage points for each $1000 increase in gross non-labour income. This softening may relate to longer-term policy changes such as the introduction and subsequent expansion of the national PPL scheme in 2011 designed to better integrate women into waged work. Evidence suggests that the scheme was effective in supporting mothers’ return to work over the longer term (Australian Government Department of Social Services, 2014).
For women employed part-time, the estimates show that education, marital status, and the presence and number of children are all positively associated with the likelihood of part-time employment (Column 3 of Table S1 in the online supplementary appendix). This aligns with existing evidence that caring responsibilities (particularly for children) remain a key driver of part-time work among Australian women (Westmore, 2024). As with full-time employment, the period interactions for non-labour income indicate that, from 2016 onwards, its effect on part-time work is essentially zero (calculated as +0.011–0.010 = 0.001). This attenuation may also reflect the longer-term impact of the PPL scheme. The remaining period interactions show no change since 2016 in the effects of education or marital status on the likelihood of part-time work.
Casual and part-time work
In her 1998 paper, Pocock observed the rapid growth of casual employment during the 1990s, noting that ‘two-thirds of those who work casually are also part-time’. Contemporary estimates indicate that this proportion has remained remarkably stable at around two-thirds (70% to be precise) (see Figure S2 in the supplementary online appendix). Among other things, the persistence of this ratio underscores the entrenched nature of part-time casual employment in Australia's labour market and the strong effect of normative factors in shaping labour market outcomes. Notwithstanding a casual loading to deter the use of casual employment, employers continue to rely heavily on part-time casuals, particularly in female-dominated industries.
Pocock (1998) also observed that, in the 1990s, many part-time workers were underemployed, that is, wanted to work more hours. Contemporary underemployment estimates are presented in Figure 3. In January 2020, prior to the start of COVID-19, 26% of women employed part-time wanted to work more hours. By April 2025 the share had dropped to 16%, consistent with a tight labour market and worker shortages. Among men the underemployment rate in the part-time labour market was 22% in April 2025. In absolute numbers there were around 452,000 women and 302,000 men in 2025 who would have preferred to work more hours. In other words, a large number of casual workers are underemployed.

The share (%) of part-time workers who are underemployed (want to work more hours), 2014 to 2025.
Segregation
Figure 4 shows total employment by industry in 2005 and 2025. Over the 20-year period total employment has grown by 5 million. Of this growth, 26.8% was in the Health, care and social assistance sector, 12.5% in Professional, scientific and technical services, 11.8% in Education and training and 11.1% in Construction. The Australian labour market, however, is highly sex-segregated which means that, at a disaggregated level, the growth patterns differ. Among women employment increased by 2.6 million over the 2005–2025 period, with 37.4% of new jobs established in the Health, care and social assistance sector, followed by 17.3% in Education and training and 10.4% in Professional, scientific and technical services. The result is little change in the industry distribution of women over the 20-year period. Indeed, some sectors such as Education and Training have become even more feminised. In 2005 66.9% of the Education and Training workforce were women. By 2025 this share had increased to 71.8%. In Health care and social assistance sector the share of women declined, marginally, from 78.7% in 2005 to 75.9% in 2025 (Figure 4).

Total employment by industry, 2005 and 2025, persons and women.
The Duncan Dissimilarity Index (DDI) provides a measure of industry or occupational segregation by quantifying the proportion of one group that would need to change for the distribution of two groups to be equal across all industries (or occupations). Using one-digit industry data, the DDI shows that in 2005 34% of women (or men) would need to change industry sectors to achieve gender parity. By 2025 the DDI was equal to 35%. In other words, the level of sex-segregation by industry was virtually unchanged. (It is important to remember that the index is only looking at horizontal segregation).
Using two-digit occupation codes the DDI also reveals persistent gender-based occupational segregation. Among full-time workers, the index in 2005 was 0.47, indicating that 47% of women (or men) in full-time employment would need to change occupations to achieve gender parity. This figure declined slightly to 0.45 in 2015 and 0.42 in 2025, suggesting a slow reduction in gender segregation at the two-digit level.
In the part-time labour market the index stood at 0.38 in 2005, falling to 0.34 in 2015, and remaining at that level in 2025. While this might suggest that the part-time labour market is less sex-segregated than the full-time labour market, it is important to remember that the index measures the distribution of men and women within part-time employment only. Since part-time work tends to be concentrated in specific occupations (e.g., retail, hospitality) the occupational distribution of men and women may appear more similar, resulting in a lower index. In other words, the lower index in the part-time labour market (vis-à-vis the full-time labour market) reflects a clustering of men and women into a narrower set of occupations, rather than there being a genuinely more integrated labour market.
When the DDI is estimated using hours worked data (all jobs) by occupation, the index was 0.42 in 2005, 0.46 in 2015 and 0.46 in 2025. Consistent with the above it suggests no change in overall horizontal occupation segregation between 2015 and 2025. This persistent clustering of men and women into specific occupations, combined with the limited opportunities typically available in the part-time labour market, reinforces gender inequalities in employment.
Figure 5 presents the occupational distribution of men and women in 2005 and 2025 using one-digit occupation data. Estimates of occupational integration based on two-digit data (not shown) indicate that some groups such as Sales Assistants and Salespersons, Numerical Clerks, and Sales and Support Workers have become more gender-integrated over time. In contrast, occupations such as Educational Professionals, Business, HR and Marketing Professionals, Health Professionals, Legal Professionals and Information and Communication Technology (ICT) Professionals have become less integrated.

Occupational distribution of men and women in 2005 and 2025.
For example, in 2005, 6.2% of women and 1.6% of men were employed as Health Professionals. By 2025, the shares had risen to 8.7% for women and 2.9% for men. Among ICT Professionals, in 2005, 0.8% of women and 2.3% of men were employed in this field. By 2025, this rose to 1.2% for women and 4.1% for men. At the two-digit occupational level the most notable decline in integration occurred among Carers and Aides. In 2005, 6.0% of women and 0.5% of men were in this occupation group. By 2025, the shares had increased to 9.4% for women and 2.3% for men. 2
Wages
Figure 6 shows trends in average weekly ordinary time earnings (AWOTE) for men and women in full-time employment, along with changes in the gender wage gap (GWG) in the full-time labour market. The conventional approach to calculating the GWG is to express the difference between men's and women's AWOTE, as a percentage of men's wages. Such an approach shows how much men's wages would need to fall for gender parity to be achieved. If the difference were expressed as a percentage of women's wages it would show how much women's wages would need to increase for gender parity to be achieved.

Wage growth and trends in the gender wage gap, November 2000 to 2024.
Between 2006 and 2017, men's AWOTE grew at a faster rate than women's. However, since 2017, this trend has reversed. Using May 2020 as a reference point, between May 2020 and November 2024, women's wages in full-time employment increased by 17.2%, compared to 14.4% for men. Over this period, the GWG has narrowed. In the private sector, it declined by 3.9 percentage points, falling to 17.2%. In the public sector, the improvement was more modest, with the gap shrinking by 0.6 percentage points to 9.3%. At the aggregate level, the GWG fell from 14.2% in May 2020 to 11.9% by November 2024.
The observed convergence may reflect factors such greater workforce attachment among women and improved access to full-time and higher paid jobs. A limitation of the Figure 6, however, is that AWOTE comparisons are only based on persons employed full-time and there is no ability to control for compositional changes in the workforce.
To overcome these limitations data from the HILDA survey along with regression analysis are used to generate and compare estimates of the GWG in the 2001–2005 period and in the more recent 2020–2023 period. The GWG is measured via a dummy variable that is equal to 1 if the respondent is a woman. The regression also includes a control for part-time status in main job of employment, thus providing a measure of the part-time full-time wage gap. Again, in the interest of space, the full set of regression results are reported in an online supplementary appendix (Table S2). The percentage effects of gender and part-time status on wages are summarised in Table 2. Two earnings measures are used. The first is hourly earnings in the main job. The second is total financial year earnings from work and salary in the previous year.
Percentage effects of gender and part-time status on wages and earnings.
Note: Employees aged 25–54, excludes full-time students. Percentage effects are calculated from exponentiated coefficients in log-linear wage regressions. All reported wage gaps are statistically significant at the 1% level. Columns (1) and (2) use hourly wages in the main job as the dependent variable, while columns (3) and (4) use financial year earnings from wages and salary. The gender wage gap is based on the coefficient on a female dummy variable. The part-time/full-time wage gap is based on the coefficient for part-time status in the main job. All regressions control for education, actual labour market experience, marital status, migrant background, number of children, geographic remoteness and state of residence and include a lambda term from a first-stage employment selection model. Full model estimates (including standard errors) are provided in Table S2 in the online supplementary appendix. Estimates are weighted to reflect population benchmarks.
Source: HILDA Survey, Waves 1–23.
The estimates in summarised in Table 2, show that in 2020–2023 women, on average, earned 10% less hourly wages than men. Part-timers earned 6.5% less than full-timers. The Column 4 results in Table 2 focus on previous financial year earnings. In 2020–2023 women, on average, had financial year earnings which were 17% less than those of her male counterpart. The part-time/full-time estimate shows the difference in previous financial year earnings of persons employed part-time or full-time at the time of the HILDA survey. In other words, it is a ‘noisier’ estimate as it could be that some part-timers worked full-time in the previous financial year or were not in the labour market. Nevertheless, as an indicative measure it shows that, at the mean, the financial year earnings of part-timers are around half that of full-time workers.
To assess whether the observed changes in wage and earnings penalties over time were statistically significant, z-tests were conducted to compare coefficients across the two periods (2001–2005 and 2020–2023). These tests examined differences in the GWG and the part-time/full-time wage differential for both hourly wages and annual earnings. Although the magnitude of the gaps changed slightly between the two periods, none of the differences were statistically significant. In other words, there has been no significant change in the GWG in mean hourly or annual earnings since 2000–05 and no change in the part-time/full-time wage gap over this period.
To check for a ‘swimming upstream’ effect a decomposition was undertaken using the Wellington approach (Wellington, 1993). The swimming upstream effect refers to the situation where women are improving their human capital (e.g., education and experience) at a faster rate than men while changes in the wage structure (the way the labour market pays for such characteristics) are working against them (e.g., returns to experience increasing at a faster rate for men) (Blau and Kahn, 1997). The decomposition approach is explained on pages 13–14 of the the supplementary online appendix. The estimates are for women and men aged 25–54. The analysis provides clear evidence of a ‘swimming upstream’ effect. Between 2001–05 and 2020–23 women′s average wages grew at a faster rate than men's, with the gender difference in growth equal to 0.016 log points. When decomposed the estimates show that 0.024 log points growth came from gender differences in characteristics (e.g., women's greater investments in education) and -0.008 from changes in the way these characteristics were paid for, with the net effect being 0.024–0.008 = 0.016 log point growth (Table S4). Table S3 in the supplementary online appendix shows that, in 2001/05 (column (5)), 31.8% of women employees in the sample held a university degree; by 2020/2023 this had risen to 52.6% (column (7)). For men, the corresponding increase was from 27.0% to 40.7%. Yet over the same period, the return to a tertiary qualification (relative to year 12) declined for women (from 35.9% to 30.9%) (columns (1) and (3)) while for men the change was negligible (from 38.1% to 37.9%) (columns (2) and (4) (These percentage effects are computed as exponentiated OLS coefficients from the results in Table S3). The changing education wage premiums may partly reflect supply-side effects: more highly qualified women in the labour market can place downward pressure on relative returns. However, they may also reflect a process of credentialism, in which employers raise formal qualification requirements without increasing the rewards associated with those qualifications. The teaching profession offers a clear example. For many years, the standard postgraduate entry route was a 1-year graduate diploma in education, but from 2014 this was replaced by a 2-year master's degree, effectively increasing the credential requirement without a corresponding improvement in pay or conditions (Joseph, 2022).
Another factor contributing to the ‘swimming upstream’ effect is the pattern of returns to labour market experience. Although average years of experience declined for both women and men between the two periods, the reduction was slightly smaller for women (a fall of 0.9 years compared with 1.2 years for men). However, the payoff to experience grew more strongly for men. By 2020/2023, each additional year of labour market experience was associated with a 3.3% increase in men's earnings, compared with a 2.3% increase for women (see Table S3).
The main take-away from this discussion is that, on average, the gender gap in pay is unchanged over 20 years. Beneath the aggregate statistics the picture is more complex. Even though women's education and work experience have improved, the way the labour market rewards (i.e., pays for) these characteristics has shifted in ways that favours men. In other words, it's not just what qualifications they hold or what experience they have that matters, but how broader economic forces such as bargaining power affects the wage structure and how, this, in turn, values particular attributes. Evidence here shows that the valuation appears to be increasingly favouring men. 3
These trends are concerning as GWGs matter. They both reflect and reinforce traditional gender norms around unpaid care work and sustain expectations about who should undertake domestic and caring responsibilities (Westmore, 2024; Pocock, 2003). They also shape how work is valued, reinforcing gendered norms about the worth of jobs in feminised occupations and sectors (Hebson and Rubery, 2018; see also Skills Australia, 2025). More broadly, GWGs contribute to economic insecurity and heighten women's vulnerability to financial dependence and gender-based violence, particularly when combined with insecure or low-paid work (Aizer, 2010; OECD 2025a).
In Australia, these inequalities in gender earnings are embedded in employment regulation and practice. For example, they are reflected in differences in award rates of pay between male-dominated and female dominated sectors that do not reflect differences in job content or skill (Austen and Preston, 2024, p 98). The privileging of the standard employment relationship of full-time permanent work has produced stronger pay and conditions in awards and enterprise agreements covering male-dominated sectors. Efforts to redress these entrenched disparities are ongoing, including, as noted, recent gender-based undervaluation cases before the FWC, such as in the FWC gender-based undervaluation priority awards review (FWC 2025c).
These structural inequities manifest not only in aggregate wage gaps but also in patterns of low pay. Women remain disproportionately concentrated in sectors and occupations where award rates are relatively low and bargaining strength is weak. Reliance on awards, limited access to collective bargaining, and employment in feminised, undervalued industries all contribute to a higher likelihood of women being in low-paid jobs. Understanding who is low paid, and why, therefore provides important insight into the institutional foundations of gender wage inequality.
Using HILDA, the likelihood of being low paid is examined by defining low pay as earning an hourly wage at or below two-thirds of the median. Estimates in Table 3 show that in 2020–2023 around 9.2% of employed women aged 25–64 were low paid, compared with 6.8% of men.
Share of employees who are low-paid (paid two-thirds or less of median hourly earnings in their main job).
Note: Employees aged 25–64.
Source: HILDA, Waves 1–23. Estimates weighted to reflect population totals.
Regression results (Table S5, online appendix) indicate that workers in Retail Trade, Accommodation and Food Services, Administrative and Support Services, and Other Services, as well as women in Health Care, have a higher probability of being in low-paid jobs than employees in manufacturing. Occupations with an elevated likelihood of low pay include Clerical and Administrative Workers, Sales Workers, Machinery Operators and Drivers, and Labourers.
Pay-setting arrangements also predict low paid employment. Relative to employees covered by collective agreements, women reliant on awards have a 5.1 percentage point higher probability of being low paid; among men, the gap is 4.1 points.
Table 4 shows persistent gender differences in how pay is set. Although the share of women reliant on awards fell from 35.2% in 2008 to 26.6% in 2023, it remains higher than for men (23.0% to 18.6%). Collective agreement coverage for women has risen slightly, but individual arrangements remain more common among men. The gender composition of the award-reliant workforce has remained stable: women consistently comprise around 60.6% of all award-covered employees, reflecting their concentration in lower-paid, award-dependent roles (see Figure S5 in the supplementary appendix).
Distribution of employees by pay-setting method and sex, various years (%).
Source: HILDA, Waves 8, 15 and 23. Sample restricted to employees aged 15–64. Estimates weighted to reflect population totals
At May 2023 total cash earnings were, at the mean, $37.7 per hour for award-only employees, $51.9 for those covered by collective agreements and $49.4 for those under individual agreements (FWC, 2025a). These differences underscore the economic implications of women's greater dependence on awards and their concentration in low-wage sectors.
Quality of working life
In this section a range of indicators from the HILDA Survey are used to assess job satisfaction and job characteristics at five intervals: 2001–2005, 2006–2010, 2011–2015, 2016–2019 and 2020–2023 (Table 5).
Measures of job satisfaction and job characteristics, by gender.
Note: Employees aged 25–64.
Source: HILDA, Waves 1–23. Estimates weighted to reflect population totals.
Across three dimensions of job quality (job security, overall job satisfaction, and flexibility to balance work and non-work commitments) there has been a clear upward trend over the past two decades. The most pronounced improvements occur in the 2020–23 period, particularly in job security satisfaction, which rises sharply for both men and women. Overall job satisfaction also increases after remaining relatively stable for many years. Satisfaction with flexibility strengthens as well, especially for men, who experience a noticeable rise in 2020–2023; women's flexibility satisfaction also increases but to a lesser extent. Women continue to report slightly higher satisfaction than men for job security and overall job satisfaction, though the gender gaps are small and narrow somewhat over time. Taken together, these outcomes point to improvements in perceived job quality in the post-2020 labour market.
Across the past two decades, workers have consistently reported relatively high levels of work intensity. Both men and women agree that they ‘have to work fast’ and ‘work very intensely’, with women reporting slightly higher levels of intensity than men in every period. These measures increase modestly for both genders between 2001–2005 and 2016–2019, before easing slightly in 2020–2023. Perceptions of not having enough time to complete required tasks show the opposite pattern: they decline somewhat for men between 2001–2005 and 2011–2015 before stabilising, while for women they increase gradually over time. Despite these shifts, gender differences remain small, with women consistently reporting a slightly greater sense of work pressure.
The availability of flexible start and finish times has increased steadily for both men and women. Among men, the proportion reporting access to this entitlement rises from 42% in 2001–2005 to 63% in 2020–2023; for women the increase is from 40% to 58%. This expansion of flexible work arrangements is consistent with broader changes in employer practices over the past decade, including the post-COVID acceleration of hybrid and flexible scheduling options.
In her 1998 review Pocock demonstrated that experiences of time pressure at work vary significantly by occupation and industry. Table 6 extends the analysis above by disaggregating job intensity (‘I have to work very intensely in my job’) by time period, gender and occupation. The data show that managers and professionals consistently report the highest levels of job intensity, followed by trades and clerical workers.
Intensity of job (‘I have to work very intensely in my job’) by occupation.
Note: Employees aged 25–64.
Source: HILDA, Waves 5–23. Estimates weighted to reflect population totals. Note the responses are on a 7-point scale: 1 = strongly disagree; 7 = strongly agree.
Across all occupations, reported job intensity has increased between 2001–05 and 2020–2023, with particularly sharp rises in lower-paid and lower-status occupations. Percentage increases are largest among women in labouring (26%) and operator roles (23%), followed by clerical and administrative workers (19%) and community and personal service workers (20%). Men in these occupations also report sizeable increases (9%–17%). Higher-skilled occupations such as managers and professionals exhibit more modest increases in intensity, ranging from 4%–5% for men and 7%–11% for women. In every occupation, women report larger percentage increases than men. These patterns indicate that job intensity has risen across the occupational structure, but the steepest increases have occurred in roles where job control is typically lower. The stronger rise among women suggests a widening gender gap in perceived work intensity. These findings are consistent with earlier evidence of gendered work intensification (Pocock, 1998) and with more recent work by Green et al. (2022), who report that women experience markedly higher levels of required work intensity than men.
Financial hardship provides another measure of wellbeing. In the following analysis, financial hardship is defined through reported difficulties in meeting essential household expenses such as being unable to pay electricity or housing costs, going without meals, or needing to pawn or sell belongings. Table 7 shows the share of employees by occupation (one digit), gender and time-period who report experiencing financial hardship. Overall (all occupations) financial hardship has declined between 2001–2005 and 2020–2023; however, the improvements have been uneven and gendered. Across all occupations and in every period, women report higher rates of hardship than men. Hardship remains most common in sales, labouring, community and personal service work, and operator roles, where between one-quarter and one-third of women continue to experience financial difficulty. While men in many of these occupations show noticeable reductions in hardship over time, particularly in trades, labouring and sales, the declines for women are smaller or non-existent, and in some cases hardship has increased (e.g., women in operator roles). Even in higher-paid occupations such as managers and professionals, women's hardship rates remain consistently above men's.
Share of employees who are experiencing financial hardship by occupation, sex and year.
Note: Employees aged 25–64. Each year participants in the HILDA Survey are asked ‘…did any of the following happen to you because of a shortage of money’ [tick all boxes]. The options are: could not pay electricity, gas or telephone bills on time; could not pay the mortgage or rent on time; pawned or sold something; went without meals; was unable to heat home; asked for financial help from friends or family; and asked for help from welfare/community organisations. If a person answered ‘yes’ to any of these questions they are defined as facing financial hardship.
Source: HILDA, Waves 1–23. Estimates weighted using the self-completion questionnaire weights to reflect population totals.
Summary and conclusion
Writing in 1998, Barbara Pocock urged policymakers to take seriously the issue of time, both the time required for unpaid care work and the degree of control workers have over their working hours. She was particularly critical of market-based solutions to gender equality, warning that such approaches ignored the interdependence of paid and unpaid work and risked further entrenching women's labour-market disadvantage.
Drawing on ABS and HILDA data, this paper examines whether women's labour-market position has improved since the early 2000s, providing an empirical update to the issues Pocock identified in the late 1990s. The evidence points to both continuity and change. For example, women's educational attainment now exceeds that of men and their participation in paid work has significantly increased, however, such developments have not produced structural shifts in their employment patterns. In September 2001, employed women worked an average of 30.3 h per week in their main job; by March 2025 this had risen only marginally to 30.4 h. This stability signals the enduring influence of gendered norms around work and care.
Occupational and industrial sex segregation has shifted only slowly, while wage gains that might have flowed from women's rising qualifications have been offset by features of the wages system that continue to advantage men (e.g., a greater share of coverage in the bargaining stream). The result is a persistent GWG. After controlling for experience, qualifications and other earnings-related factors, the adjusted gender gap in mean hourly wages remains around 10% in 2020–2023, with no statistically significant change since 2001–2005.
Recent government reforms offer some grounds for optimism. The planned extension of PPL to 26 weeks by July 2026 may help shift norms around care and increased subsidies for early childhood education and care are likely to ease constraints on women's participation when they have young children (Australian Government, 2024). Reforms to workplace rights, including expanded eligibility to request flexible working arrangements due to pregnancy or family and domestic violence, stronger obligations on employers to genuinely consider flexible work requests, and new powers for the FWC to arbitrate disputes, all represent further progress. Nonetheless, calls remain for stronger protections, including universal access to access to the right to request flexible working arrangements and limiting employer refusals to cases of ‘unjustifiable hardship’ rather than ‘reasonable business grounds’ (Smith and Charlesworth, 2024, p.26).
Policy must also continue to address gender-based undervaluation. In this context, the FWC's work on identifying and correcting gender-based undervaluation through a series of cases focusing on feminised occupations within feminised awards is vital. However, the process is resource-intensive, uneven, and slow to deliver outcomes. A further concern is that any improvements in the GWG achieved through work-value adjustments in feminised modern awards may be eroded if wage growth in male-dominated enterprise-bargaining streams grows more rapidly. This issue was recognised in the recent review of the 2022 SJBP Act, which proposed several reforms (Recommendations 10–13), including that the Australian Government ‘actively monitor wage-setting practices, especially in enterprise agreements, to ensure that modern award outcomes lead to sustained improvements in gender pay equity’ (Bray and Preston, 2025, p. 230).
Recent national benchmarking reinforces these findings. The University of Sydney's new Gender Equality @ Work Index (Hill et al. 2025), which tracks trends across participation, pay, hours, job quality, segmentation, security and safety from 2014–2024, reports only modest improvement in gender equality at work over the past decade. In the decade to 2024 the overall index rose by three points to 83 (an index of 100 suggests parity based on the methodology in Hill et al.) Although the index uses a different methodology to this paper, its findings align with the broader patterns identified here, namely incremental progress alongside persistent structural barriers.
Taken together, the evidence in this paper points to the continued relevance of the structural forces identified by Pocock more than two decades ago. Although women's educational attainment and labour-force participation have increased, progress remains constrained by gendered norms around care, limited access to secure and predictable hours and the undervaluation of feminised work. Without sustained institutional and policy attention to these constraints, improvements in women's labour-market outcomes are likely to remain slow and uneven.
Looking ahead, the emerging productivity debate poses new challenges and risk for women, especially as it is likely to be influenced by orthodox economics. 4 Labour productivity has fallen below trend, with highly feminised sectors such as Health, care and social assistance now being described as a ‘drag on productivity’ (FWC, 2025b, PN209; Maltman, 2024). This framing shifts responsibility onto care workers rather than acknowledging the structural undervaluation of their work and the difficulties of measuring productivity in care-intensive sectors (Bray and Preston, 2025, appendix 4). The potential consequences were illustrated in the 2025 Annual Wage Review, where weak labour productivity growth was cited as a ‘restraining factor’ on the size of the increase that the FWC settled on (FWC, 2025, para.9).
These developments underscore the importance of labour-market policies that recognise the intersection of paid and unpaid work and highlight the need for caution when relying on orthodox economics to guide labour-market decision making. Correcting gender inequalities is not only essential for fairness but also vital to sustaining economic performance and social equity, particularly as demographic pressures and care demands intensify. Ensuring that women can participate fully in secure, predictable and properly valued work is integral to Australia's long-term economic prosperity (Skills Australia, 2025).
While this paper has focused primarily on gender, broader intersectional analyses remain important for understanding how other characteristics such as race, migration status, disability and age, and interact with gender to shape labour-market outcomes. Future research that brings these intersecting dimensions into view will be important to developing policy approaches that are both effective and equitable.
Supplemental Material
sj-docx-1-jir-10.1177_00221856261440564 - Supplemental material for All change, still gendered: The Australian labour market in the 2020s
Supplemental material, sj-docx-1-jir-10.1177_00221856261440564 for All change, still gendered: The Australian labour market in the 2020s by Alison Preston in Journal of Industrial Relations
Footnotes
1.
It is noteworthy that, in the last decade, the hours spent on unpaid adult care grew faster than the hours spent on unpaid childcare (see Figure S3 in the supplementary appendix). For an analysis of gender differences in time spent on employment, housework and care see, also,
.
2.
To shed further light on changing employment patterns
in the supplementary online appendix focuses on employment changes within the health, care and social assistance sector. The chart shows the new jobs have gone in terms of one-digit occupation and hours worked (full-time, part-time). Over the 10-year period of 2015 to 2025 total employment in this sector increased by 59% or 855,800 new jobs. Much of this growth (73%) has been since 2019. Figure S4 shows that between 2015–2019 Professionals (full-time plus part-time) occupied half (51%) of the new jobs, with a further 23% of the new jobs going to Community and Professional Service Workers. This group includes Health and welfare support workers, Carers and Aides. Between 2019 and 2025, of the new jobs created, Professionals have moved into 40% of them and 43% have gone to Community and Personal Service Workers.
3.
The analysis presented here focuses on the GWG at the mean. While there is growing interest in understanding wage outcomes across the distribution (given that wage-setting processes may differ for sub-groups such as low- and high-skilled workers), a full distributional analysis is beyond the scope of this paper. Such an extension would require more complex decomposition techniques and lies outside the empirical focus adopted here. Relevant work does exist, however.
provide a decomposition of the GWG for young adults in Australia between 2001–2002 and 2018–2019. Consistent with the findings reported in this paper, they observe no significant change in the gap over time. They also find that the GWG is larger at the top of the wage distribution than at the median in both periods, concluding that ‘breaking down the glass ceiling is, therefore, as hard today as it was at the turn of this century’ (p. 381).
Acknowledgments
The paper uses data from the Household, Income and Labour Dynamics in Australia (HILDA) survey. I acknowledge that the HILDA Project was initiated and is funded by the Australian Government Department of Social Services (DSS) and is managed by the Melbourne Institute of Applied Economic and Social Research (Melbourne Institute). The findings and views reported in this paper, however, are those of the author and should not be attributed to either the DSS or the Melbourne Institute.
Author note
An earlier version of this paper was presented to the New South Wales Industrial Relations Society, Newcastle Branch, and I thank participants for their comments and suggestions. I would like to thank two anonymous reviewers for the Journal of Industrial Relations for their helpful comments and suggestions. I am especially grateful to Emerita Professor Sara Charlesworth for her invaluable input, particularly during the revision of the manuscript. All remaining errors and omissions are my own.
Funding
The author received no financial support for the research, authorship and/or publication of this article.
Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Disclaimer
OpenAI's ChatGPT was used to assist with language editing. No generative artificial intelligence was used for data analysis. The analysis, findings and views reported in this paper are entirely the author's own.
Supplemental material
Supplemental material for this article is available online.
Author biography
Alison Preston is a Professor of Economics in the Department of Economics within the UWA Business School at the University of Western Australia.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
