Inflexibility and the stalled progress of American women in the labour market

The labour force participation of women in the US peaked in 1997 and has stagnated since. This column argues that workplace inflexibility is the common thread running through this puzzle, using COVID-19 as a natural experiment. Women were hit harder than men in the initial downturn due to both over-representation in high-contact and inflexible service occupations and school closures. When remote work relaxed flexibility constraints, mothers’ employment rebounded strongly above pre-pandemic levels. But as return-to-office mandates have intensified, occupational inflexibility has re-emerged as a significant predictor of employment declines – and only for women.

The Trump administration’s 2025 order returning federal employees to the office five days a week has crystallised a debate that has been building across corporate America. Amazon, Disney, Google, JPMorgan, and a long list of other large employers have rescinded the hybrid policies they extended during the pandemic and reasserted in-office attendance as the default. Commercial real-estate data show that the share of large firms requiring full-time in-office work climbed from roughly 5% in 2023 to about 54% in 2025 (JLL 2025), and survey work by Barrero et al. (2023) and follow-up data from the Survey of Working Arrangements and Attitudes (Barrero et al. 2025) document a parallel hardening of employer expectations. Some chief executives argue, often pointedly, that this is good for women’s careers. The evidence assembled in Albanesi (2026) – and a now substantial literature on remote work, family policy, and women’s labour supply – suggests the opposite.

The puzzle of stalled progress

Goldin’s (2006, 2014, 2021) sweeping work on the ‘quiet revolution’ documents how American women transformed their relationship with the labour market over the 20th century. Yet that progress stalled. Female labour force participation in the US plateaued near 59% in 1997 and never resumed its earlier upward march. In a cross-country comparison of prime-age (25–54) participation rates among 22 advanced economies, the US slid from 5th place in 1990 to 21st in 2019 (OECD 2024). Blau and Kahn (2013) attribute a large share of this divergence to the gap between the US and other rich countries in family policies. Albanesi et al. (2023) extend that analysis and show that the US invests significantly less in family policies than comparable countries across virtually every margin.

Inflexibility as the unifying concept

Why has the US fallen behind? The common thread linking the leading candidate explanations is workplace inflexibility – the degree to which jobs require physical presence, fixed schedules, and continuous attachment to the labour market. This inflexibility interacts with care responsibilities, which continue to fall disproportionately on women, to bind their labour supply.

The first manifestation lies in the structure of earnings. Goldin and Katz (2011) and Goldin (2014) document large penalties for reduced hours and any kind of workplace flexibility in professional and managerial occupations – the very occupations into which college-educated women were moving in growing numbers from the 1980s onward. The 1990s saw an accelerated rise in top earnings among college-educated men (Heathcote et al. 2010) and a widening gender wage gap among college workers (Albanesi and Prados 2022). Albanesi and Prados (2022) find that changes in the earnings structure can account for roughly one-third of the slowdown in participation for women married to college-educated men, generating both a negative income effect through their husbands’ rising earnings and stronger frictions to their own career advancement.

The second manifestation is the absence of supportive family policies. Since the Family and Medical Leave Act of 1993, the US has made no federal progress on paid leave, public childcare, or workplace flexibility regulation. The US ranks last among 17 OECD comparators on the OECD’s part-time generosity index, has among the lowest cash benefits for families as a share of GDP, and is the only country in the sample without any paid federal parental leave (Albanesi et al. 2023). A simple cross-country counterfactual reported in Albanesi (2026) suggests that adopting Sweden’s policy mix would raise the female/male employment ratio for married workers in the US by about 2.5 percentage points – modest, but meaningful in a labour market where the participation gap has been all but frozen for three decades.

COVID-19 as a natural experiment

The pandemic provided a sudden, largely exogenous relaxation of flexibility constraints. Bloom and colleagues tracked this shift in real time (Aksoy et al. 2024, Shah et al. 2024), showing that working from home stabilised at roughly four times its pre-pandemic rate by mid-2023. Before COVID-19, the principal route to flexibility for workers with care responsibilities was to reduce hours – and to absorb the steep wage penalty that came with it. Remote-work technology fundamentally altered this trade-off: workers could maintain high hours while gaining flexibility over when and where those hours were performed, effectively decoupling flexibility from the hours penalty embedded in the earnings structure.

Albanesi and Kim (2021) showed that women were hit harder than men in the initial 2020 downturn, both because they were over-represented in high-contact and inflexible service occupations and because school closures placed binding care constraints on mothers. The subsequent recovery, however, was striking. Figure 1 plots the change in employment for prime-age women relative to 2019 by family status, controlling for age, education, and occupation. The deep 2020 trough is followed by a rebound that takes married and single mothers from a 2.5 percentage-point shortfall back to roughly one percentage point above 2019 levels by 2023 – a striking reversal of the longer-run stagnation. 

Figure 1 Change in employment for prime-age (25–54) women relative to 2019, by family status

Notes: Bars report annual averages of same-month differences in the employment rate, conditional on age, educational attainment, and two-digit SOC occupation. Error bars are 95% confidence intervals. ‘With children’ denotes co-residence with a child under 12. Note the strong recovery through 2023 and the visible retrenchment in 2024–2025 for women with children.
Source: Author’s calculations from the Current Population Survey, Bureau of Labor Statistics; see Albanesi (2026) for details.

Causal evidence reinforces the pattern: in a randomised controlled trial at a large technology firm, Bloom et al. (2024) find that hybrid work reduced quit rates by about one-third, with the reduction concentrated among women and workers with long commutes. Survey and revealed-preference work places a high monetary value on the remote-work option (Mas and Pallais 2017, Cullen et al. 2026), and Aksoy et al. (2025) show that fully remote firms have disproportionately expanded the hiring of married women and workers outside major cities.

The return-to-office reversal

From late 2023 onward, employer-provided flexibility has been rolled back. Pew Research Center (2024) reports that the share of US workers required in the office regularly rose from about 63% in early 2023 to roughly 75% in late 2024; JLL (2025) finds that 54% of large firms now require full-time on-site work, up from about 5% two years earlier. Hybrid arrangements remain widespread (Gallup 2025), but formal mandates have intensified.

The right-hand side of Figure 1 makes the consequences visible: from 2023 to 2025, employment falls back toward 2019 levels for every demographic group, and the drop is sharpest for mothers. Single mothers’ employment declines by roughly 0.9 percentage points and married mothers’ by about 0.5 percentage points relative to 2023. 

To examine whether this retrenchment is linked to the rollback of flexibility, I estimate cross-sectional regressions in Albanesi (2026) of employment change on occupational inflexibility – measured from O*NET task descriptors as the degree to which a job requires physical presence and fixed schedules (Albanesi and Kim 2021) – separately by gender, across two windows: the 2019–2023 recovery and the 2023–2025 return-to-office period. Between 2019 and 2023, employment grew significantly more in flexible occupations for both men and women: a one-standard-deviation increase in inflexibility is associated with roughly 9 percentage points lower employment growth for each gender. Between 2023 and 2025, however, the relationship becomes gender-specific. For men, inflexibility no longer predicts employment changes at all. For women, it continues to predict significant declines.

This asymmetry is hard to attribute to demand-side factors. Women are disproportionately employed in service occupations, which expanded over this period. Women’s employment is also typically less cyclically sensitive than men’s (Albanesi 2025), so the appearance of inflexibility effects for women but not men represents a reversal of the usual gender pattern in labour-market fluctuations. Selection is also unlikely to drive the result: if women with weaker attachment had sorted out of inflexible occupations during the pandemic, the remaining female workforce in those occupations would be positively selected, biasing against the finding. The most parsimonious reading is that the withdrawal of flexibility reimposed binding constraints on women’s labour supply – exactly the mechanism the conceptual framework predicts.

Implications for policy

Three implications follow. First, public policy in this domain still matters, perhaps more than ever. Where employer-provided flexibility is being rescinded, the case for federal action on childcare, paid parental leave, predictable scheduling, and the tax treatment of secondary earners is strengthened, not weakened. 

Second, remote work is not a substitute for family policy. It is concentrated among college-educated workers in professional and managerial occupations; for workers without a college degree, the binding flexibility constraint is more often unpredictable scheduling in service jobs. Davis et al. (2026) further note that remote work, while genuinely valued, can only partially offset the structural forces depressing both fertility and women’s employment in countries with weak family-policy infrastructure. 

Third, the post-2023 reversal illustrates that employer-provided flexibility is unstable. A decentralised market arrangement that depends on management discretion is vulnerable to peer effects, executive turnover, and changing cultural fashion in a way that statutory entitlements are not.

American women’s labour-market progress did not stall by accident. It stalled in an institutional environment that combined a particularly steep hours-earnings gradient with an unusually thin layer of public support for families. The pandemic briefly loosened both constraints. Whether the gains of that period prove permanent depends on choices – by firms and by legislators – that are being made right now.

Source : VOXeu