Gender differences in work expectations and psychological distress: insights from the United States and South Korea

Abstract

Objectives:

Over the past 50 years, the roles of women, particularly those from the “Baby Boom” cohort, have shifted from primarily domestic responsibilities to active participation in the labor force. Moreover, differences in gender norms and sociocultural contexts across countries influence women’s labor force participation. This study examined gender differences in the association between work expectations and psychological distress among Baby Boomers in the United States (1946–1964) and South Korea (1955–1963).

Methods:

Data came from the 2006–2018 waves of the Health and Retirement Study (n = 14,005) and the Korean Longitudinal Study of Aging (n = 2,362). Perceived expectations of working in the next 5 years were reported on a probability scale (0–100%). Psychological distress was assessed by the Center for Epidemiologic Studies Depression Scale. Multivariate mixed-effects logistic regressions were used to examine the longitudinal association between work expectations and psychological distress for men and women.

Results:

In both countries, women had higher CES-D scores and lower expectations of working in the next 5 years than men. Higher work expectations were robustly associated with lower odds of psychological distress among both men and women in Korea; however, this inverse association was more modest in the United States. The association between expectations and distress was stronger among Korean men (OR_men = 0.83 [0.79, 0.87]) compared to Korean women (OR_women = 0.89 [0.84, 0.94]); there were no gender differences among U.S. respondents (OR_men = 0.98 [0.94, 1.02]; OR_women = 0.98 [0.96, 1.00]).

Discussion:

Findings highlight the complex interplay between biological sex and social contexts in shaping expectations regarding work in later life.

Among the most recognizable social movements in the 20th century, the women’s rights movement significantly transformed the lives of the “Baby Boom” generation (born 1946–1964), a cohort who entered adulthood in the late 1960s and early 1970s (Agree, 2017). The movement empowered this generation of women to defy traditional gender attitudes in an era when their primary social expectations were to be the homemakers and caregivers of their families. Women from the Baby Boom generation had unprecedented access to educational and employment opportunities. They are the first to achieve higher rates of college completion than men (Bailey & DiPrete, 2016). Since the 1960s, women of this cohort have also entered the workforce in historic numbers. In the United States (U.S.), the labor force participation rate of the Baby Boom women increased dramatically, from 38% in 1960 to 43% in 1970, and reached 60% by 2000 (Toossi, 2002). These educational and employment gains have brought a notable shift in the roles of women in society.

Implications of women’s labor force participation for employment in later life

Participation in the workforce has both positive and negative implications for working women as they age. On the positive side, work provides personal fulfillment and greater economic freedom as women become more financially independent (Richardson, 1999). For example, retired women from the Baby Boom generation are among the first who receive pension incomes based on their own earnings; in prior generations, pension benefits were often limited to married women as spousal benefits based on their partners’ earnings. In 2014, the Social Security Administration reported that more than 70% of Baby Boom women received retirement benefits based on their earnings rather than marriage-based benefits; this figure was estimated at 60% for older cohorts (Iams, 2016). However, despite increasing access to paid work outside the home, women’s responsibilities in the domain of family life have been slower to change. Among married couples where both members work, women are still the primary caregivers of the household (Pacheco Barzallo et al., 2024), and women spend more time on housework and caregiving activities than men (Langner & Furstenberg, 2020). They are more likely to take leave from work or take longer leaves for child and elder family care (Herr et al., 2020). Thus, women often juggle multiple roles as employees, spouses, mothers, and caregivers (Spurlock, 1995).

Among dual-earner households, men and women navigate work and family responsibilities differently to balance demands from work and family roles (Noor, 2004), which subsequently influences their employment behaviors. Men’s workforce participation typically follows a linear employment trajectory (e.g., full-time employment to complete retirement) (Richardson, 1999). In contrast, women may transition between various career paths, from full-time employment to intermittent work such as part-time employment or exiting the workforce for child or elder care responsibilities (Damaske & Frech, 2016). This gendered pattern of work trajectories likely arises from the differences in employment preferences between men and women (Jost & Möser, 2023; Wiswall & Zafar, 2018). Indeed, gender plays an important role in shaping the expectations regarding late-life employment (Abrams et al., 2022). These expectations reflect older adults’ preferences about their future employment and are shaped by individual characteristics (e.g., gender, race) and factors beyond the individuals (e.g., culture, gender norms). For example, women generally report lower expectations of working later in life than men (Abrams et al., 2022), and women of racial/ethnic minorities were less likely to plan to delay retirement (McNamara & Williamson, 2004). Education and social class are also correlated with expectations, with higher expectations of working longer among individuals with more educational attainment and higher social class (Fisher et al., 2016).

Moreover, contexts of employment, such as the life history of the Baby Boom cohort and gender norms regarding workforce participation, further contribute to gender differences in work expectations. The Baby Boom cohort entered the workforce during a period of economic prosperity when an abundance of employment opportunities contributed to an increase in labor force participation, particularly among workers from this cohort (Poulos & Nightingale, 1997). While both genders of this cohort have strong attachments to the workforce (Fry, 2019), labor force participation is strongly dependent on gender norms and social expectations for men’s and women’s roles. In particular, while average workers retire between age 62 and 67 (Fisher et al., 2016), women typically retire at younger ages to assume caregiving duties, a pattern that is more evident in a collectivist culture (Kim, 2022). These findings suggest the need to situate the effect of gender on employment in a broader societal context, particularly among the Baby Boom women who experienced significant shifts in gender norms as they often balance both work and family responsibilities throughout adulthood.

Gender norms and employment expectations in a cross-country context

The earlier discussion regarding the Baby Boom generation primarily focuses on the employment context and gender norms in the United States; however, sociocultural and historical differences across countries also influence employment decisions among older adults. These contextual features include retirement norms (e.g., mandatory retirement practice), cultural values (e.g., individualism versus collectivism), gender norms (e.g., pro-women’s rights and gender equality), and workplace policies (e.g., anti-discrimination laws).

Retirement norms vary considerably across countries. In many Asian countries, such as South Korea, mandatory retirement practices enforce or incentivize older workers to retire involuntarily at a specified age limit, as early as mid-50s. In contrast, mandatory retirement is less common in countries such as the United States, and most workers typically retire in their early 60s (OECD, 2024). Countries also vary in cultural characteristics; among them, the individualism-collectivism dimension is particularly salient to this study. Individualistic cultures, such as the United States, place emphasis on individual interests and achievements over a social group; in contrast, collectivist cultures in many Asian nations prioritize the needs of a social group over an individual and family cohesion and responsibilities. These cultural dimensions influence how individuals experience and navigate work and family responsibilities within a given culture. For example, workers in individualistic cultures report more work-interference-with-family conflicts (Spector et al., 2007), whereas those in collectivist cultures report more family-interference-with-work conflicts (Yang et al., 2000). Thus, individuals in collectivist cultures may be more interested in part-time employment as it allows them to continue to fulfill their family responsibilities (Oishi et al., 2015). In addition, gender norms also shape workforce participation for men and women. For instance, in Asian countries, more entrenched gender inequality norms may result in discriminatory employment practices that limit women’s opportunities in the workforce (Cislaghi et al., 2022). Finally, workplace and family-related policies, including equal pay and antidiscrimination laws, can influence employment decisions, especially for female workers (Aycan, 2008; Bailey & DiPrete, 2016). These findings reflect the social and cultural differences in work/family roles across countries, suggesting opportunities for cross-country comparisons to gain insights into the contextual factors that shape employment expectations in later life.

Conceptual framework linking work expectations to mental health

A large and diverse body of research has explored how subjective assessments of the future (e.g., expectations of major transitions, such as the likelihood of retiring, moving into a nursing home, living to a certain age, as well as overall impressions of the future, such as optimism and hopelessness) relate to mental health (Miloyan et al., 2017; Tang et al., 2024; Thimm et al., 2013). Indeed, work expectations are longitudinally associated with symptoms of psychological distress among older adults (Abrams et al., 2022; Mezuk et al., 2022). Specifically, low work expectations (e.g., reporting <10% probability of working in the future) are positively associated with feelings of distress among older adults (Mezuk et al., 2022). Research on the psychological correlates of expectations is consistent with theories about the causes of poor mental health, which emphasize the roles of hopelessness and perceived uncertainty about the future (Liu et al., 2015; Saulnier et al., 2019). Expectations of major late-life transitions, such as retirement, are particularly salient for mental health because these transitional periods often entail changes in social roles and health behaviors (Dang et al., 2022). Thus, expectations regarding work may relate to psychological distress via correlations with hopelessness and intolerance to uncertainty in the context of life events (Saulnier et al., 2019; Thimm et al., 2013).

Furthermore, the association between work expectations and mental health may differ by gender because men and women are predisposed to different workplace stressors (Biswas et al., 2021). As a prime example, the wage gap is a widely known gender inequality relating to employment, with women earning significantly less than men even when both work for the same employer or in the same occupation (Penner et al., 2023). This gender wage gap increases with age; for example, American women ages 55–64 earn only 80% as much as men of the same ages, compared to a ratio of 92% among workers ages 25–34 (Kochhar, 2023). Gender inequalities in employment opportunities (e.g., higher prevalence of unpaid or part-time work among women) also contribute to a higher risk of psychological distress among women compared to men (Gueltzow et al., 2023). Lastly, workplace exposures differ by gender. Specifically, men are more likely to work in physically demanding occupations and at a higher risk for workplace injuries, whereas women are more likely to experience discrimination and harassment, which may elevate their risk for poor mental health (Biswas et al., 2021). Over the life course, these differences contribute to differences in work and mental health of older men and women (Ervin et al., 2022).

Present study

While previous research showed that the subjective expectations regarding future employment were associated with depressive symptomatology among older adults (Abrams et al., 2022; Mezuk et al., 2022), it is unclear whether this association differs for men and women. Thus, this study examined gender differences in the association between work expectations and psychological distress among the “Baby Boom” cohort who experienced a unique labor force history (i.e., increasing proportion of women in the workforce). Moreover, the association between expectations and psychological distress was explored in the United States and South Korea (“Korea” thereafter). This cross-country comparison is informed by key differences in contextual characteristics relating to gender across countries, including cultural norms surrounding gender roles and workplace policies; however, we acknowledge that variation in gender-related experiences within a country (e.g., by race/ethnicity and socioeconomic status) may also shape the association between work expectations and mental health. Based on our prior findings (Mezuk et al., 2022), we hypothesized that higher work expectations are associated with a lower likelihood of psychological distress. Further, the inverse association between expectations and distress will be stronger among men because men often place a stronger emphasis on work roles than women. Lastly, the gender differences in the association between work expectations and psychological distress will differ between the United States and Korea because of contextual differences relating to gender across countries.

Method

Data source and study population

Data came from two population-based longitudinal studies of aging conducted in the United States and Korea: the Health and Retirement Study (HRS, n~20,000 respondents ages 51+ surveyed biennially since 1992) and the Korean Longitudinal Study of Aging (KLoSA, n~10,000 respondents ages 45+ surveyed biennially since 2006). Both the HRS and KLoSA employ a complex, multi-stage study design that enrolls refreshment cohorts (every 6 years in the HRS and at the 2014 wave in KLoSA) to remain representative of the contemporaneous older adult population. These harmonized studies provide a unique opportunity to investigate the association between the expectations for late-life employment and psychological distress among older men and women in the United States and Korea. Additional details on the HRS and KLoSA are described elsewhere (Joon et al., 2007; Sonnega et al., 2014).

This study focused on the Baby Boom cohorts born in the United States (birth years: 1948–1965) and Korea (birth years: 1955–1963). The years 1948–1965 were used as the birth years of the U.S. Baby Boomers because they reflect the HRS sampling design. We selected different years of birth in the KLoSA and HRS to account for the historical background of the Baby Boom cohorts in two countries. Specifically, the Korean Baby Boom followed the Korean War (1950–1953) and occurred later than the start of the U.S. Baby Boom, which followed World War II (1939–1945). As a result, the KLoSA sample is 5-years younger than the HRS sample, and the analyses adjusted for several age-related covariates to account for the confounding effect of age (see “Statistical approach”).

The full analytical sample included 16,367 non-proxy respondents (14,005 Americans and 2,362 Koreans) who had complete data on work expectations, psychological distress, and all potential covariates for at least one interview wave from 2006 to 2018. These interview waves were chosen because 2006 was the first interview wave of KLoSA, and 2018 was the latest available interview wave that was not impacted by the COVID-19 pandemic. Given the assessment of work expectations in KLoSA, the Korean sample, by design, included only respondents who were working at any point between 2006 and 2018. In addition to the full sample, a sample of 10,502 U.S. respondents who were working for at least one interview wave between 2006 and 2018 was also selected for the main analysis to better align with the Korean sample. Supplementary Figure 1 (see online supplementary material) details the analytic sample selection process. Differences between the full analytic sample and the entire sample of Baby Boomers who were interviewed between 2006 and 2018 but excluded due to missing data on key variables of interest are summarized in Supplementary Table 1 (see online supplementary material). Most relevant to this investigation, the excluded respondents were less likely to be working compared to the full analytic sample.

The HRS and KLoSA are approved by the research ethics review committees in their respective countries, and all respondents are provided written informed consent. This analysis used only publicly available data (https://g2aging.org) and was exempt from human subject regulation.

Work expectations

Questions on work expectations were asked at every interview wave in both the HRS and KLoSA; however, there are several key differences between the two datasets. First, the KLoSA assessed work expectations by a single question asking the respondents about their subjective likelihood of working “at the present job in the next 5 years.” Specifically, respondents younger than 50 were asked about the likelihood of working after age 55; respondents aged 50–54 were asked about working after age 60; respondents aged 55–59 were asked about working after age 65; and finally, respondents aged 60+ were asked about the likelihood of working in the next 5 years. On the other hand, the HRS assessed work expectations using three questions that asked the respondents about the subjective likelihood of “working full-time” after age 62, 65, and 70. These questions were then used to create a new variable based on respondents’ age: expectations of working after age 62 for respondents aged 56 or younger; expectations of working after age 65 for those aged 57–62; and expectations of working after age 70 for those older than age 62. Overall, this variable assessed the average likelihood of working full-time over the next 5 years. Second, responses were reported on a probability scale of 0 to 100 in 10-point intervals in KLoSA, whereas responses were on a continuous scale from 0 to 100 in the HRS. Finally, work expectations were assessed only among respondents who were working at the time of assessment in the KLoSA; in contrast, the HRS assessed work expectations in all respondents irrespective of their working status. This difference in the assessment of work expectations informed the basis of the sensitivity analyses described below.

Work expectations were first modeled as a continuous time-varying exposure, and responses were rescaled to 0 to 10 for a more meaningful interpretation of regression coefficients (i.e., a one-unit increase in the scaled work expectation is equivalent to a ten-unit increase in the original variable). Additionally, reflecting the multi-modal distribution of this variable (e.g., clusters at tens and at 0%, 50% and 100%), a categorical version of work expectation (operationalized as 0%–10%, 11%–40%, 41%–60%, 61%–89%, 90%–100%) to explore potential non-linear association between work expectations and psychological distress, with the category of 41%–60% used as the reference to reflect “50/50 chance” uncertainty.

Psychological distress

Psychological distress was assessed by the Center for Epidemiologic Studies Depression Scale (CES-D) on all respondents at every interview wave. The CES-D is originally a 20-item screening instrument that assesses four domains associated with distress symptoms (i.e., depressed affect, positive affect, somatic symptoms, and interpersonal relations) during the past week (Radloff, 1977). Although the HRS and KLoSA both derive their psychological distress measures from the CES-D, the HRS uses an 8-item scale, each recorded as yes/no (CES-D 8; scores range from 0 to 8). In contrast, the KLoSA uses a 10-item scale, each recorded on a 4-point Likert scale ranging from “very rarely (less than 1 day)” to “almost always (5 to 7 days)” (CES-D 10; scores range from 0 to 30). Specific items in CES-D 8 and CES-D 10 are summarized in Supplementary Table 2 (see online supplementary material). CES-D 8 and CES-D 10 have comparable psychometric properties (O’Halloran et al., 2014), acceptable internal consistency and test-retest reliability in older adults (Cronbach alpha is 0.78–0.84 for CES-D 8 and 0.72–0.80 for CES-D 10) (Cheng & Chan, 2005; Dang et al., 2020).

The CES-D summary scores were constructed by summing all symptoms, reverse-coding items on positive effects (e.g., “happiness” and “enjoying life”); higher scores indicated more severe symptomatology. To facilitate the comparison between the HRS and KLoSA, summary scores were then dichotomized to classify cases of psychological distress using threshold scores of ≥4 symptoms for CES-D 8 (HRS) and ≥10 for CES-D 10 (KLoSA). Previous research shows that these thresholds are equivalent to a threshold of 16 or more symptoms on the original 20-item CES-D (Steffick, 2000) and demonstrate acceptable sensitivity and specificity for distress symptoms (Cheng & Chan, 2005; Dang et al., 2020).

Covariates

All models were adjusted for demographic and health covariates that have been associated with work expectations and psychological distress in prior work. Demographics included age at baseline (centered at each sample’s mean), years of follow-up, education (less than lower secondary education, upper secondary & vocational training, tertiary education), marital status (married, not married), minority status, household income, total wealth, and employment status. Minority status (yes/no) was classified as a racial/ethnic minority group of the United States (e.g., non-Hispanic Black, Hispanics, and others) and rural residential status for Korea. This variable indicates racialized or marginalized groups that are disadvantaged in access to resources, healthcare, and employment opportunities, allowing for a common adjustment across countries. Household income and total wealth were categorized into quartiles using baseline data. Employment status was categorized into full-time, part-time, self-employed, unemployed, partly retired, retired, disabled, and not in the labor force. Health characteristics included self-rated health (poor/fair, good, very good/excellent) and medical morbidity, indicated by having at least one chronic condition among hypertension, diabetes, cancer, chronic lung disease, heart problem, stroke, and arthritis. All covariates were time-varying as a function of interview wave, except for race/ethnicity and baseline age.

Statistical approach

Initially, baseline demographic and health characteristics for men and women were compared using Wilcoxon rank sum tests (comparing means) and Kruskal–Wallis tests (comparing medians) for continuous variables and chi-squared tests for categorical variables.

Mixed effects logistic regressions with random intercept were used to examine the association between work expectations and psychological distress over the 12-year period (see Supplementary Methods for details, see online supplementary material). This modeling approach accounts for the clustered structure in the data (i.e., within-person over time) and the time-varying nature of work expectations and psychological distress. It also incorporates information from all assessments rather than only those observations with complete follow-up over the study period (Hardin & Hilbe, 2018). The inclusion of a random intercept term accounted for heterogeneity in baseline psychological distress; the likelihood ratio test of the random-intercept model versus the model without was highly significant (p < .0001). A fixed effect for years of follow-up was included to allow each respondent to have a different length of follow-up time. We also fitted models with both random intercept and random slope for years of follow-up to allow for different rates of change in distress status over time, and overall, found similar results to models with random intercept only (Supplementary Table 3, see online supplementary material). In addition, models with interaction terms between years of follow-up and baseline age (to test whether the rate of change in psychological distress over time depends on baseline age) as well as between years of follow-up and work expectations (to test whether the association between work expectations and psychological distress changed over time) were explored; however, these interaction terms were not statistically significant and thus removed from the analysis.

All models were fit separately for men and women and by country (the United States, Korea). For each country, differences in the association between work expectations and psychological distress across men and women were tested by including an interaction term between self-reported gender and work expectations in the full sample of both men and women. Analyses using the HRS were conducted in the full sample as well as among current U.S. workers to better align with the assessment of work expectations in KLoSA. Plots of predicted average marginal probabilities for psychological distress were displayed to visualize the findings.

Statistical analyses were performed in the R environment, version 4.2.2 (2022). Significance tests were evaluated two-sided at p < .05. Mixed effects logistic regressions were fit via glmer, which estimates maximum likelihood by Laplace approximation. Marginal mean predicted probabilities were estimated with the R package ggeffects, version 1.5.0.

Sensitivity analyses

A series of sensitivity analyses was conducted to assess the robustness of the results. Since there is no consensus on CES-D threshold scores among older adults, sensitivity analyses were conducted using alternative thresholds (≥3, ≥5 symptoms for CES-D 8 and ≥11, ≥12 symptoms for CES-D 10) to classify cases of psychological distress (Cheng & Chan, 2005; Dang et al., 2020). Additional analyses using the HRS were also conducted among current U.S. workers who were born between 1955 and 1963 to address the age difference between the United States and Korean respondents.

Results

Descriptive statistics

Table 1 shows the characteristics of U.S. and Korean respondents at the interview wave when they first entered the study during the period between 2006 and 2018. In both countries, women had higher CES-D scores and lower mean expectations of working in the next 5 years than men. They were also less likely to be married, had lower household income and wealth, and were less likely to be working (p < .05). In Korea, women were more likely to have lower education than men; in contrast, the opposite pattern was observed in the United States, where women had higher educational attainment.

Table 1.

Sample characteristics of the U.S. and Korean respondents at the interview wave when they first entered the study during the study period between 2006 and 2018.

Variable	United States						Korea
	Full sample			Sample of respondents who were working for at least one wave			Full sample
	Men (n = 6,174)	Women (n = 7,831)	p a	Men (n = 4,869)	Women (n = 5,633)	p a	Men (n = 1,216)	Women (n = 1,146)	p a
Psychological distress
Normalized CES-D (mean ± SD)b	0.20 ± 0.25	0.24 ± 0.29	<.0001	0.15 ± 0.22	0.19 ± 0.25	<.0001	0.16 ± 0.13	0.17 ± 0.14	0.1
CES-D 8 ≥ 4 or CES-D 10 ≥ 10, N (%)c	964 (16.0)	1,679 (21.6)	<.0001	510 (10.7)	878 (15.7)	<.0001	111 (9.2)	148 (12.9)	.004
Work expectations
Mean ± SD	45.0 ± 37.4	36.6 ± 36.7	<.0001	53.2 ± 35.5	45.9 ± 36.3	<.0001	82.4 ± 19.9	78.2 ± 22.7	<.0001
Median (interquartile range)	50 (80)	25 (75)	<.0001	50 (60)	50 (70)	<.0001	80 (30)	80 (30)	<.0001
Age (mean ± SD)	54.4 ± 3.0	53.8 ± 3.5	<.0001	54.2 ± 2.8	53.6 ± 3.3	<.0001	48.9 ± 2.7	48.5 ± 2.3	.1
Minority group, N (%)	3,095 (50.1)	4,075 (52.0)	.03	2,299 (47.2)	2,751 (48.8)	0.1	172 (14.1)	191 (16.7)	.1
Education, N (%)
Lower secondary	1,011 (16.4)	1,173 (15.0)		636 (13.1)	562 (10.0)		211 (17.4)	399 (34.8)
Upper secondary, vocational training	3,623 (58.7)	4,751 (60.7)	.03	2,821 (57.9)	3,427 (60.8)	<.0001	653 (53.7)	641 (55.9)	<.0001
Tertiary	1,540 (24.9)	1,907 (24.4)		1,412 (29.0)	1,644 (29.2)		352 (28.9)	106 (9.2)
Marital status, N (%)
Never married	545 (8.8)	719 (9.2)		344 (7.1)	449 (8.0)		34 (2.8)	14 (1.2)
Married/partnered	4,589 (74.4)	4,896 (62.6)	<.0001	3,816 (78.4)	3,652 (64.9)	<.0001	1,127 (92.7)	1,004 (87.6)	<.0001
Divorced/separated	930 (15.1)	1,791 (22.9)		647 (13.3)	1,263 (22.4)		44 (3.6)	63 (5.5)
Widowed	106 (1.7)	427 (5.5)		60 (1.2)	267 (4.7)		11 (0.9)	65 (5.7)
Household income quartiles, N (%) d
First (lowest)	1,354 (21.9)	2,150 (27.5)		650 (13.3)	963 (17.1)		265 (21.8)	372 (32.5)
Second	1,470 (23.8)	2,029 (25.9)	<.0001	1,140 (23.4)	1,533 (27.2)	<.0001	286 (23.5)	258 (22.5)	<.0001
Third	1,613 (26.1)	1,889 (24.1)		1,444 (29.7)	1,591 (28.2)		336 (27.6)	256 (22.4)
Fourth (highest)	1,737 (28.1)	1,793 (22.5)		1,635 (33.6)	1,546 (27.4)		329 (27.1)	259 (22.6)
Total wealth quartiles, N (%) e
First (poorest)	1,427 (23.1)	2,094 (26.7)		830 (17.0)	1,128 (20.0)		272 (23.7)	299 (27.2)
Second	1,558 (25.2)	1,924 (24.6)	<.0001	1,225 (25.2)	1,413 (25.1)	.0004	269 (23.4)	284 (25.8)	.01
Third	1,582 (25.6)	1,919 (24.5)		1,368 (28.1)	1,555 (27.6)		291 (25.3)	272 (24.7)
Fourth (richest)	1,607 (26.0)	1,894 (24.2)		1,446 (29.7)	1,537 (27.3)		316 (27.5)	244 (22.2)
Employment status, N (%)
Full-time	3,269 (52.9)	3,458 (44.2)		3,237 (66.5)	3,408 (60.5)		474 (40.2)	253 (22.7)
Part-time	288 (4.7)	828 (10.6)		281 (5.8)	807 (14.3)		53 (4.5)	103 (9.2)
Self-employed	871 (14.1)	646 (8.2)		856 (17.6)	632 (11.2)		589 (49.9)	297 (26.7)
Unpaid family work for 18+ hoursf	–	–	<.0001	–	–	<.0001	5 (0.4)	115 (10.3)	<.0001
Unemployed	419 (6.8)	457 (5.8)		214 (4.4)	227 (4.0)		22 (1.9)	27 (2.4)
Partly retired/retired	818 (13.2)	1,164 (14.9)		150 (3.1)	265 (4.7)		10 (0.8)	50 (4.5)
Disabled/not in labor force	509 (8.2)	1,278 (16.3)		131 (2.7)	294 (5.2)		27 (2.3)	269 (24.2)
At least one condition: hypertension, diabetes, cancer, or chronic lung disease	3,982 (64.5)	5,334 (68.1)	<.0001	2,912 (59.8)	3,531 (62.7)	.003	223 (18.3)	199 (17.4)	.6

Note. CES-D = Center for Epidemiologic Studies - Depression Scale.

^aWilcoxon rank sum tests (comparing means) and Kruskal–Wallis tests (comparing medians) were used for continuous variables; chi-squared tests were used for categorical variables.

^bThe normalized CES-D score for each respondent was calculated as the sum of all symptom scores divided by the maximum possible score, ranging from 0 to 1.

^cMinority status indicated a racial/ethnic minority group for the United States (e.g., non-Hispanic Black, Hispanics, and others) and rural residential status for Korea.

^dHousehold income quartiles in U.S. respondents (USD): <21,994, 21,994–51,480, 51,480–100,476, ≥100,476; Korean respondents (10,000 Korean won): <1,203, 1,203–2,586, 2,586–4,404, ≥4,404.

^eTotal wealth quartiles in U.S. respondents (USD): <3,005, 3,005–76,000, 76,000–291,000, ≥291,000; Korean respondents (10,000 Korean won): <1,532, 1,532–7,000, 7,000–16,155, ≥16,155.

^fThe employment category “unpaid family work for 18+ hours” was only in KLoSA.

Figure 1 shows the mean work expectations for men and women, stratified by psychological distress status, in the United States and Korea from 2006 to 2018. Overall, Korean respondents reported higher mean work expectations than U.S. respondents. In both countries, mean work expectations did not substantially change over the study period, and lower expectations were observed among respondents experiencing psychological distress.

Figure 1.

Mean work expectations of Baby Boom men and women by psychological distress status in (A) the United States (N = 14,005) and (B) Korea (N = 2,362) from 2006 to 2018. Psychological distress was defined as having CES-D 8 ≥ 4 (for U.S. respondents) or CES-D 10 ≥ 10 (for Korean respondents). Work expectations were assessed on a continuous probability scale, ranging from 0% to 100%. Mean work expectations were calculated conditional on the psychological distress status at each follow-up wave. CES-D = Center for Epidemiologic Studies - Depression Scale.

Continuous work expectations and psychological distress among men and women

As shown in Table 2, in both countries, higher expectations of working in the next 5 years were associated with lower odds of psychological distress after adjusting for demographic and health characteristics. However, this inverse association was of a weaker magnitude in the United States. Moreover, despite having the same magnitude, the inverse association between expectations and distress was statistically significant among women (odds ratio (OR)_women = 0.98, 95% confidence interval (CI): 0.96, 1.00 for every 10% increase in work expectations) but not men (OR_men = 0.98 [95%CI = 0.94, 1.02]) in the United States. Interaction test indicated that this gender difference was not statistically significant (p = .07) (however, it may still be theoretically important despite the lack of statistical significance). In Korea, the inverse association was significantly stronger among men (OR_men = 0.83 [95%CI = 0.79, 0.87) compared with women (OR_women = 0.89 [95%CI = 0.84, 0.94]; p = .04). Figure 2 illustrates these findings, showing that, in the United States, higher work expectations were associated with slightly smaller predicted probabilities of psychological distress among women, but no associations between work expectations and psychological distress were observed among men (Figure 2A). In Korea, work expectations were also inversely associated with psychological distress, and this association was more pronounced in men than in women (Figure 2B).

Table 2.

Odd ratios and 95% confidence intervals for the association between continuous work expectations and psychological distress ^a by gender in the United States and Korea.

Variable	United States			South Koreab
	Men	Women	p c	Men	Women	p c
Full sample, n	6,174	7,831		1,216	1,146
Effect estimate	0.99 (0.96, 1.01)	0.96 (0.95, 0.98)*	.21	0.83 (0.79, 0.87)*	0.89 (0.84, 0.94)*	0.04
Among respondents working for at least one wave, n	4,869	5,633
Effect estimate	0.98 (0.94, 1.02)	0.98 (0.96, 1.00)*	.07

Note. CES-D = Center for Epidemiologic Studies - Depression Scale; KLoSA = Korean Longitudinal Studies of Aging.

^aPsychological distress was defined as having CES-D 8 ≥ 4 (the United States) or CES-D 10 ≥ 10 (Korea). Work expectations were modeled as a continuous variable and scaled to 0–10. Models included random intercept, centered baseline age, years of follow-up, minority status, education, marital status, household income, total wealth, self-rated health, having at least one chronic condition, and employment status.

^bThe Korean sample, by design, included only respondents who were working at any point between 2006 and 2018 because of the way work expectations were assessed in KLoSA.

^cThe p-value corresponds to the interaction effect between self-reported gender and work expectations on psychological distress in the model.

^*p < .05.

Figure 2.

Average marginal predicted probability of psychological distress for men and women who were working for at least one interview wave between 2006 and 2018 in (A) the United States and (B) Korea. Average marginal predicted probabilities for having CES-D 8 ≥ 4 (the United States) or CES-D 10 ≥ 10 (Korea) were estimated using continuous work expectations (range: 0%–100%). Models included random intercept, centered baseline age, years of follow-up, minority status, education, marital status, household income, total wealth, self-rated health, having at least one chronic condition, and employment status. CES-D = Center for Epidemiologic Studies - Depression Scale.

Categorical work expectations and psychological distress in men and women

Next, to explore potential non-linear relationships, the exposure of work expectations was modeled as a categorical variable, with the reference group of “uncertain” (expectations of 41%–60%). Table 3 shows that, in the United States, using this categorization, there was no significant association between work expectations and psychological distress among men; however, women who reported low expectations of working (i.e., 0%–10%) had significantly higher odds of distress relative to those who were uncertain (OR_women = 1.23 [95%CI = 1.00, 1.50]). In Korea, both men and women with lower expectations of working generally had higher odds for psychological distress, though this inverse association was only statistically significant when comparing respondents who reported expectations of 11%–40% with those with uncertain expectations (OR_men = 1.53 [95%CI = 1.03, 2.27] and OR_women = 1.62 [95%CI = 1.07, 2.45]). Additionally, higher expectations of working (between 61% and 100%) were associated with lower odds of distress in both Korean men (OR_61%–89% = 0.66 [95%CI = 0.50, 0.86 and OR_90%–100% = 0.44 [95%CI = 0.32, 0.61]) and women (OR_61%–89% = 0.65 [95%CI = 0.48, 0.88] and OR_90%–100% = 0.66 [95%CI = 0.47, 0.93]).

Table 3.

Odd ratios and 95% confidence intervals for the association between categorical work expectations and psychological distress ^a by gender in the United States and Korea.

Variable	United States		South Koreab
	Men	Women	Men	Women
Full sample, n	6,174	7,831	1,216	1,146
0–10%	1.11 (0.90, 1.38)	1.29 (1.10. 1.51)*	1.39 (0.67, 2.89)	0.98 (0.43, 2.26)
11–40%	0.90 (0.70, 1.16)	0.99 (0.82, 1.19)	1.53 (1.03, 2.27)*	1.62 (1.07, 2.45)*
41–60%	Ref	Ref	Ref	Ref
61–89%	0.81 (0.63, 1.05)	0.96 (0.79, 1.18)	0.66 (0.50, 0.86)*	0.65 (0.48, 0.88)*
90–100%	1.12 (0.87, 1.45)	0.96 (0.79, 1.18)	0.44 (0.32, 0.61)*	0.66 (0.47, 0.93)*
Among respondents working for at least one wave, n	4,869	5,633
0–10%	1.16 (0.83, 1.63)	1.23 (1.00, 1.50)*
11–40%	0.85 (0.59, 1.23)	0.89 (0.70, 1.13)
41–60%	Ref	Ref
61–89%	0.71 (0.49, 1.03)	0.96 (0.76, 1.22)
90–100%	1.16 (0.80, 1.68)	0.99 (0.78, 1.26)

Note. CES-D = Center for Epidemiologic Studies - Depression Scale; KLoSA = Korean Longitudinal Study of Aging; Ref = Reference group.

^aPsychological distress was defined as having CES-D 8 ≥ 4 (the United States) or CES-D 10 ≥ 10 (Korea). Models included random intercept, centered baseline age, years of follow-up, minority status, education, marital status, household income, total wealth, self-rated health, having at least one chronic condition, and employment status.

^bThe Korean sample, by design, included only respondents who were working at any point between 2006 and 2018 because of the way work expectations were assessed in KLoSA.

^*p < .05.

Sensitivity analyses

As shown in Supplementary Tables 4 and 5 (see online supplementary material), using alternative threshold scores of CES-D (≥3, ≥5 symptoms for CES-D 8 and ≥11, ≥12 symptoms for CES-D 10) for classifying psychological distress similarly found a stronger inverse association between work expectations and psychological distress among Korean men compared to women, although the difference between men and women was slightly smaller than that of the main analysis. In the United States, results were consistent in that the inverse association between work expectations and psychological distress was statistically significant among women but not men, but no gender differences were found. Similar results were found when the analyses in the HRS were restricted to current U.S. workers with years of birth between 1955 and 1963 (Supplementary Table 6, see online supplementary material).

Discussion

This longitudinal study explored the relationships between gender, work, and mental health among the Baby Boom generation using an international comparative approach. There are three major findings: First, higher work expectations were robustly associated with lower odds of psychological distress among both men and women of the Baby Boom generation. Second, this inverse association was more modest among U.S. respondents compared to their Korean counterparts. Finally, the inverse association between work expectations and psychological distress was stronger among Korean men compared to women, but there were no gender differences among U.S. respondents. Collectively, these findings suggest that there are multifaceted processes shaping gender differences in the association between work expectations and psychological distress among Baby Boomers. This study illustrates the ways in which contextual factors, including age cohort, gender, and social norms, contribute to the links between work and mental health in later life.

In Korea, the robust inverse association between work expectations and psychological distress may reflect the employment practices and pension policies that are salient to the mental health of older workers. One example of this is the finding that older Korean adults, overall, expect to work longer than their U.S. counterparts (Table 1). The expectation of working longer could be driven by financial necessity as a consequence of a weak social safety net. In Korea, the Basic Pension system provides pension benefits for older adults with earnings in the bottom 70% of the income distribution. However, the gross pension replacement rates are relatively low; in 2022, this rate was estimated at 31% for a full-career worker [from age 22 to 59] with an average wage compared to an international average of 50% (OECD, 2019). Moreover, the current pension system largely excludes non-regular employment such as self-employed, part-time, or temporary work. These types of non-regular forms of work do not typically offer pension benefits or pay contributions to public pensions (Kuhlmann & Nullmeier, 2022). However, these work arrangements are especially prevalent among older Koreans, accounting for nearly half of salaried workers over age 60 (OECD, 2019). Given the widespread practice of “honorary” retirement (in which Korean firms incentivize older workers to retire, often in the mid-50s age range), many older Koreans involuntarily retire early from their career jobs before they are qualified to receive pension benefits [at age 60] (Cho & Kim, 2005). These older adults often seek new employment in non-regular jobs to earn or supplement their living expenses (OECD, 2019). In the context of such a precarious work environment, the association between work expectations and distress could reflect the distress around securing stable employment among older Koreans.

In the United States, work expectations were only modestly inversely associated with psychological distress among women. While overall, the inverse association between expectations and distress is consistent with findings from our prior study examining work expectations and indicators of depression in the HRS (Mezuk et al., 2022), the magnitude is smaller. This difference in magnitude may stem from differences in measurement: in the present study, psychological distress was measured using the CES-D, which only assesses current (i.e., past week) symptoms. In contrast, our prior study used the Composite International Diagnostic Interview (CIDI), a fully structured interview based on the Diagnostic and Statistical Manual of Mental Disorders-IV criteria for major depressive episodes, to assess depressive symptoms over the past 12 months. Given that planning for future employment or an eventual exit from the workforce typically takes place over an extended period of months or years (Topa et al., 2018), the CIDI may be better suited to assessing psychological distress associated with long-term retirement planning. Furthermore, it is possible that the effect of gender on the expectation-distress association is more complex and may be influenced by factors such as family/spousal relationships (Fisher et al., 2016). This hypothesis is supported by findings from previous studies on this topic; for example, a recent analysis found that marriage/partnership and gender roles moderated the association between employment stability and mental health (Cortès-Franch et al., 2018).

Exploring cohort differences using a cross-national approach

The key finding from this study is that while the association between work expectations and psychological distress was stronger among Korean men compared to Korean women, there was no significant gender difference among their U.S. counterparts. This cross-country finding should be interpreted from the life course perspective, which emphasizes how temporal (e.g., experiences earlier in life) and contextual (e.g., societal norms about gender roles) factors shape work and mental health in later life. The Baby Boom cohort entered the workforce when both countries underwent significant economic expansion following World-War II (U.S. Baby Boomers) and the Korean War (Korean Baby Boomers) (Park, 2019; Pruitt, 2020). These periods of economic growth created a thriving employment environment that encouraged a greater labor force participation, especially among male and female workers from the Baby Boom cohort (Poulos & Nightingale, 1997). However, labor force participation among women is significantly shaped by the social norms regarding gender roles within each culture. Compared to the United States, Korean culture is influenced by Confucian values, which maintain a rigid gender norm and clear division of labor for men and women (Fauser & Kim, 2023). These cross-cultural differences in gender norms influence the selection and retention in the workforce for men and women in the United States and Korea. For example, Korean women are more likely to exit the labor force for caregiving responsibilities or engage in non-regular forms of work after childbirth than men (Kim, 2022). Indeed, data from the Korean Labor and Income Panel Study showed that the majority of Korean women held more precarious, non-regular work positions, such as unpaid family work, self-employment, and low pay, temporary jobs; only one-fifth were employed in regular jobs after childbirth (Fauser & Kim, 2023). It is possible that the gendered patterns in the association between work expectations and psychological distress in the United States and Korea may reflect more entrenched gender norms in Korean Confucian culture, in which men may place a greater emphasis on work roles.

Overall, findings illustrate complex processes that contribute to the retirement process among men and women. Epidemiologic research in the area of work and health has long recognized the distinction between sex (i.e., as a biological construct based on genetics, anatomical and physiological differences) and gender (i.e., a social construct based on societal and cultural norms for men and women) (Smith & Koehoorn, 2016). Such a distinction seeks to differentiate the biological aspect from the social and cultural aspects of gender. However, sex and gender often interact to influence behaviors, including retirement decisions (Springer et al., 2012). This study provides further evidence to support such interaction, showing that the differences in the association between work expectations and psychological distress varied not only by biological sex (male/female) but also by country contexts, which guide the societal expectations for men’s and women’s roles. It emphasizes the importance of integrating both biological and social risk factors to inform mental health efforts supporting older men and women during the retirement transition.

Findings should be interpreted considering study limitations and strengths. First, the main analyses were conducted among currently working respondents. Therefore, findings may not extend to older adults who were not currently working (e.g., unemployed, temporarily laid off). These employment transitions are more prevalent among women, who may work intermittently or take time off work for caregiving responsibilities (Richardson, 1999). Second, possible reverse causation may confound results given that psychological distress could affect the expectations and ability to work. Future studies could address the potential of reverse causality by including data on early life experiences and prior depression diagnoses. Third, the study did not have data on gender attitudes and individual perceptions of work and family responsibilities. Such data is particularly important in cross-cultural contexts where work may have different meanings in Korean and American cultures (e.g., a financial means to support family vs. career achievement). Fourth, psychological distress was assessed by the CES-D rather than clinical interviews, and thus may not reflect clinically significant diagnoses. Future studies could use more robust depression measures or examine other mental health outcomes (e.g., anxiety) and related outcomes (e.g., loneliness). Lastly, the analytical data predates the COVID-19 pandemic, and the study did not examine the impact of the pandemic on the association between work expectations and psychological distress.

Despite limitations, this study has several strengths. To the best of our knowledge, this study is among the first to examine gender differences in the association between work expectations and psychological distress among older adults. While most studies on work and mental health have been conducted in Western countries, this study situated the association between work expectations and distress within the United States and Korea. This cross-country comparison provides important insights into how societal norms and expectations about gender roles influence individuals’ decisions about work and their impacts on psychological distress. Future research can extend this analysis to other countries to provide a more comprehensive, international perspective of how work expectations relate to psychological distress among men and women. Insights from such comparisons help to inform culturally appropriate interventions to support the mental health needs of an increasingly diverse population of older workers. In addition, it is important to recognize that despite harmonization, the HRS and KLoSA were conducted among different populations and countries with distinctive cultural, social, and historical contexts. As such, the study conducted several additional analyses to address the contextual differences while allowing for a more comparable comparison across countries. For example, we conducted a sensitivity analysis that restricted the HRS sample to current U.S. workers with the same years of birth as Korean workers as a way to address the age difference between the United States and Korean samples. Finally, the study sample came from large population-based studies with rich data on employment and distress measures, which enhances the study’s reliability.

Conclusion

The study found an inverse association between work expectations and psychological distress among both men and women of the Baby Boom generation. This inverse association was stronger among Korean men compared to Korean women, but there were no gender differences among U.S. respondents. Findings highlight the multifaceted processes, including age cohort, gender, and social norms, that shape work and mental health in later life.

Supplementary Material

Supplementary data are available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online.

Data availability

The HRS and KLoSA are publicly available to researchers at https://hrs.isr.umich.edu and https://survey.keis.or.kr/eng/klosa/klosa01.jsp