Rising young worker despair in the United States

Abstract

Between the early 1990s and 2015 the relationship between mental despair and age was hump-shaped in the United States: it rose in middle-age, then declined later in life. That relationship has now changed: mental despair declines with age due to a rise in despair among the young. However, we show for the first time that the relationship between age and mental despair differs by labor market status. The hump-shape in age still exists among non-workers. The change in the age-despair profile over time is due to increasing despair among young workers, a trend that began in 2010. Prior to that, the young were less likely to be in despair if in work, but today they are as likely to be in despair as students and the unemployed. We find broad-based evidence for this finding in the Behavioral Risk Factor Surveillance System (BRFSS) of 1993–2025, the National Survey on Drug Use and Health (NSDUH), 2008–2023, and in surveys by Pew, the Conference Board and Johns Hopkins University.

Highlights

• •Between the early 1990s and 2015 the relationship between mental despair and age was hump-shaped in the United States.
• •That relationship has now changed: mental despair declines monotonically with age due to a rise in despair among the young. However, the relationship between age and mental despair differs by labor market status.
• •The hump-shape in age still exists for those who are unable to work and the unemployed. The relation between mental despair and age is broadly flat, and has remained so, for homemakers, students and the retired. The change in the age-despair profile over time is due to increasing despair among young workers.
• •Whilst the relationship between mental despair and age has always been downward sloping among workers, this relationship has become more pronounced due to a rise in mental despair among young workers.

1. Declining well-being of young workers

There is a growing literature suggesting that the U-shape in age in well-being and the hump shape in age in ill-being have both disappeared in the years since around 2013–2015 in the United States and the United Kingdom (Blanchflower et al., 2024) and in Western Europe as well as in Latin America, the Middle East and Asia, although evidence is less apparent for Africa (Blanchflower & Bryson, 2025a-e). This shift has been driven by the decline in the mental health of the young and is particularly evident for young people aged below 25, and especially young women (see Haidt, 2024; Twenge et al., 2018; Udupa et al., 2023). Worsening youth mental health has subsequently been recorded in Iceland (Thorisdottir et al., 2021), Norway (Krokstad et al., 2022), Canada (Garriguet, 2021; Huang et al., 2025) and Australia (Leigh & Robson, 2025). Indeed, Blanchflower (2025) found that the young were the unhappiest age group in 167 United Nations countries. The decline in the mental health of the young is particularly evident among young women (Haidt, 2024; Twenge et al., 2018; Udupa et al., 2023).

There is also evidence of rising mental health hospitalizations of the young (Arakelyan et al., 2023; Leyenaar et al., 2025; Yard et al., 2021), rising youth suicide rates (Ehlman et al., 2022; Garnett & Curtin, 2023; Leigh & Robson, 2025; Marcotte & Hansen, 2023), and rising drug overdose deaths among the young in the United States (Miller et al., 2025; Tanz et al., 2022).

In this paper, for the first time, we report that the declining mental health of the young in the United States is driven by a decline in the mental health of young workers. This has happened in the years since 2015. We make use of data from the Behavioral Risk Factor Survey System (BRFSS) files of 1993–2025 using measures of 'negative affect' (despair) and a cognitive/evaluative measure (life satisfaction). This builds on earlier work where we showed an overall worsening of youth mental health and a collapse in both the hump-shape in ill-being and the U-shape in well-being in age (Blanchflower et al., 2025; Twenge and Blanchflower, 2025).

The existing literature suggests potential explanations for the rise in mental ill-health among the young. The one receiving most attention has been the rise in social media and access to the internet (Haidt, 2024; Suárez Álvarez & Vicente, 2024; Lohman, 2015). A series of natural experiments summarized by Pugno (2024) suggest the relationship between worsening youth mental health and the spread of the internet is causal. Nevertheless, even if social media and internet usage is a contributory factor, it is unlikely to be the only explanation of the change.

It is usual in the well-being literature to consider the role played by economic factors, including paid employment. Broadly speaking, those in paid employment tend to be happier than those who are unemployed and seeking work, and those who are incapable of work (Winkelmann & Winkelmann, 1998). Economists tend to put this down to the benefits of a higher income that comes with paid employment, even if this relationship is non-linear (Kahneman & Deaton, 2010). However, some economists have demonstrated that the value of paid work, in terms of subjective well-being, exceeds that which is due to its pecuniary value alone (Blanchflower & Oswald, 2011). This point is taken as a given for most psychologists who maintain that individuals’ self-worth, their identity and their social standing are often linked, and sometimes determined, by their employment and occupational status (Faunce, 1989). Furthermore, job loss is a shock to workers, resulting in a decline in their well-being which exceeds that which could be accounted for by the income loss (Layard et al., 2012) and which, unlike most other shocks, they are unable to recover from until they are able to find work again (Clark et al., 2008). Notwithstanding these well-known findings in the literature, no research to date has considered the role that paid work might play in the changing age profile of despair in the United States.

It is possible that changes to jobs and the labor market might lead to a reduction in the wellbeing returns to paid employment. Green et al. (2024) argue that job quality is on a par with health as a key determinant of subjective wellbeing, so changes in that job quality can be expected to impact worker wellbeing. A slowdown in the rate of real earnings growth in recent years has, arguably, reduced the overall value of paid work, or perhaps the extent to which it compensates workers for the effort exerted in working. An alternative way of looking at this is to consider time-variance in the value of leisure time: recent studies note that the price of leisure has fallen due to technological changes pushing down the price of items that young people in particular use in their leisure time, such as gaming devices (Kopytov et al., 2023). In addition, by pushing up the quality of leisure time, these trends raise the relative cost of time in employment.

Earned income, and the time it takes to make that income, do not fully capture the costs and benefits of paid employment. Psychologists from Maslow (1943) onwards, emphasize the importance to human beings of satisfying a hierarchy of needs, beginning with physiological and social needs, through to self-actualization at the top of the hierarchy. Although employment is often considered a means to achieve safety and security, relative to reliance on state welfare or family and friends, changes in employment contracts may have limited the value of paid work for some in this regard. The advent of ‘gig’ working, for example, whilst beneficial for some, may create uncertainty and insecurity for those who resort to it because they have no clear path to permanent employment and career progression. These new forms of employment are predominantly undertaken by young people (Lepanjuuri et al., 2018).

More broadly, employers are successfully deploying new technologies to minimize ‘break’ times, and exert greater control over production processes, often aided by close technological monitoring of work processes, which limit worker control and autonomy over ever-more-demanding processes, all of which – based on Karasek's (1979) theory regarding the importance of worker control and autonomy for wellbeing – should result in a decline in the wellbeing of workers. Evidence from task-based studies of work, and social surveys in which workers report on the nature of job tasks, indicates there has been a growth in job demands and a reduction in worker job control in the United Kingdom (Green et al., 2022) which, presumably, is mirrored in the United States. During COVID, the shift to home and hybrid working, whilst beneficial in some respects, may have exacerbated feelings of social isolation experienced by the young in particular as they missed out on the social component of the workplace. The demise of collective bargaining and trade union presence in the workplace implies a diminution in workers' bargaining power, making it even more difficult for workers to resist such changes and to alter their terms and conditions of employment (Feiveson, 2023).

It is plausible that the trends described above have particularly adverse consequences for young employees because they are new arrivals in the labor market and, as such, are more likely to be subject to new forms of contracting and are less likely to work in unionized workplaces where workers might draw on collective bargaining power to resist employer-driven change. Laura Feiveson (2024) from the US Treasury has evaluated the worsening, relative position of young people and argues that there are several recent negative factors lowering the well-being of the young, including young workers:

“… many changes have contributed to an increasing sense of economic fragility among young adults. Young male labor force participation has dropped significantly over the past thirty years, and young male earnings have stagnated, particularly for workers with less education. The relative prices of housing and childcare have risen. Average student debt per person has risen sharply, weighing down household balance sheets and contributing to a delay in household formation. The health of young adults has deteriorated, as seen in increases in social isolation, obesity, and death rates … An aging population means that young adults today are competing for houses and jobs with more older workers than the young of their parents' generation did. Increased globalization and technological advances brought an abundance of affordable goods to American consumers, but with the cost of fewer job opportunities for men without college degrees”.

These trends in the world of work may be accentuated by the impact of social media. As Pugno (2025) suggests, social media may exacerbate workers’ feelings of distress and despair where workers end up comparing their situation with that of others:

“As far as social competition is concerned, it is well known that comparing one’s economic condition with that of others deflates subjective well-being, at least in advanced countries, especially if the comparison is made with those with better conditions. The reason is that looking at others' economic conditions gives rise to material aspirations that are unrealistic and hence frustrating” (2024, p.7).

To our knowledge we are the first to explore the link between labor force status and the shift in the age-pattern of mental despair from a hump-shape to a monotonic decline in age. We do so with data for the United States. We find the hump-shape in age still exists for non-workers. The change in the age-despair profile over time is due to increasing despair among young workers. Whilst the relationship between mental despair and age has always been downward sloping among workers, this relationship has become more pronounced due to a rise in mental despair among young workers.

In Section Two we describe our data and methodology. In Section Three we present our Results before concluding in Section Four.

2. Data and methods

Our primary analysis uses annual micro-data for the period 1993–2025 from the Behavioral Risk Factor Surveillance System (BRFSS). Each survey contains some responses from the succeeding year. For example, the 2024 file which has a total of 457,670 observations contains 19,553 observations for January and February 2025.1 We also examine confirmatory evidence on mental health from the National Survey on Drug Use and Health (NSDUH) 2008–2023 again for the USA. Our focus is on young people aged below 25, but we include older people in the analysis as a comparator group. We include people over working age but are cautious about the interpretation of what happens to well-being after the age of 65, in part due to the positive impact subjective well-being has on surviving into older age.2

We examine changes in well-being over time by age and gender. We present simple distributions in graphical and tabular form as well as regression adjusted estimates which enter age as a categorical variable, alongside a female dummy variable and year and state fixed effects.

Our initial dependent variable is the number of bad mental health days (MHD) in the last month which is defined as follows.

Q1. “Now thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?”

We then construct our mental despair measure (Q2) by setting the Q1 variable to one when an individual gave the answer 30 and zero otherwise. This variable was initially defined and used in Blanchflower and Oswald (2020) who showed its incidence had risen over time, from 3.6 % in 1993 to 6.4 % in 2019. Blanchflower (2020) noted the rise, and the high level in this variable among native Americans. We update the Blanchflower and Oswald (2020) analysis by examining an additional four surveys from 2020 to 2024 that include the COVID shutdown plus a few months from 2025. It is clear the observed changes started before COVID, which simply extended a pre-existing trend.3

The overall weighted mean for mental distress is 5.3 % over the years 1993–2025, with a standard deviation of .224 and a sample size of over ten million (n = 10,516,969). Throughout we focus on differences over time in the relationship between age and despair for workers and non-workers. In the years 2020–2025 56.2 % of the sample was employed; 6.2 % were unemployed; 5.1 % were homemakers, 4.9 % were students, 20 % were retired and 6.2 % were ‘unable to work’ and 1.4 % refused to answer the labor force status question (n = 1,721,923).

As background, adult unemployment rates and youth unemployment rates have tracked each other over the last decade and been relatively low, but with a markedly higher youth rate of around two and a half times.4 For example, at the time of writing, the latest data for May 2025 showed an unemployment rate for those age 25–54 of 3.6 % versus 9.7 % for those age 18–24, 2.85 times higher. The recent rise in youth rates has been greater than adult rates from a low of 3.3 % to a high of 3.6 % for adults and from 7.3 % to 9.7 % for the young.

We also examine life satisfaction in the BRFSS

Q3. In general, how satisfied are you with your life? 1 = very dissatisfied, 2 = dissatisfied, 3 = satisfied 4 = very satisfied.

This question was asked of all respondents in the years 2005–2010 and was then asked in a small subset of states in the years through 2018.5 There are no observations for the years 2019–2021 but from 2022 it was then asked in about half of the states, with the distribution by states reported in Appendix 1. All states plus the British Virgin Islands, Puerto Rico and Guam were covered in the years 2005–2010. The period since then has more limited coverage by year, with six states having no coverage – Michigan, Nebraska, Ohio, Oregon, New York and Pennsylvania.

In total there are 2,406,122 observations in the years 2005–2018 (mean = 3.39, SD = .62) and 678,748 observations in the years 2022–2025 (weighted mean = 3.35, SD = .63). Examining young workers aged 18–24 below (weighted) for four representative years where we have large samples, we see that there is little difference between the average life satisfaction scores of young women and young men. Life satisfaction for both declined over time. The proportion reporting being in despair is always higher for women and rose for young men and women but more so for women.

	Life satisfaction		Despair
	Women	Men	Women	Men
2005	3.33	3.32	4.9	3.6
2010	3.38	3.38	4.4	2.5
2022	3.17	3.17	11.6	7.6
2024	3.18	3.18	9.3	6.1

3. Results

3.1. Evidence on declining mental health and life satisfaction among young workers from the BRFSS

Fig. 1 presents the age profile of despair over time for the population. The y-axis shows the share of individuals in despair. We see the hump shaped age distribution gets more pronounced over time through to 2015. But then the left-hand side for those under age 40 starts to rise from 2016 to 2019, a movement that becomes more pronounced for 2020–2025. The function then essentially declines in age. By contrast, there is little change over time in despair for those aged over fifty. Thus, from 2020, the overall hump-shape has gone. Furthermore, despair peaks when people are in their early-50s until the last period when the peak shifts to age 23, at which age around 8.5 percent of individuals are in despair in 2020–2025.

Fig. 1.

Despair by age 1993–2025, USA.

The distributions and means are reported in Table 1 for bad mental health days for five time periods: 1993–1999, 2000–2009, 2010–2015, 2016–2019 and 2020–2025. The final column shows the distribution in 2020–2025 for workers only. The median, and the mode, are zero in every period. However, the proportion saying zero days has declined over time from over two-thirds (68.5 percent) to 58.3 percent in 2020–2025. The percentage in despair (those for whom 30 out of the last 30 days were bad mental health days) has risen from 4.2 percent to 6.6 percent over the period. Despair is a little lower in the most recent period among workers – 5.4 percent say 30 of the last 30 days were bad mental health days.

Table 1.

Distribution # bad mental health days (weighted by _llcpwt)).

	1993–1999	2000–2009	2010–2015	2016–2019	2020–2025	2020–2025 (workers)
0	68.5	66.0	65.5	64.1	58.3	57.3
1	3.5	3.7	3.4	3.3	3.2	3.5
2	6.0	5.9	5.3	5.1	5.4	6.0
3	3.3	3.4	3.2	3.3	3.7	4.1
4-29	18.7	21.0	22.6	24.2	22.8	23.7
30	4.2	5.8	5.6	6.0	6.6	5.4
Mean	2.95	3.38	3.73	4.01	4.76	4.36
N	873,937	2,978,477	2,758,189	1,767,288	2,139,078	1,078,050

Table 2 takes the last period and examines variance in both the number of bad mental health days and despair for different sub-populations. The number of bad mental health days and mental despair incidence are higher for women than men and decline in age. Mental health also varies markedly by race: natives have especially high rates of mental despair, whereas Asians have low rates. Both the number of bad mental health days and mental despair are higher among those unable to work, as compared to those in work, but mental despair is lower among the unemployed than it is among employees and the self-employed. Following Kuroki (2021), we define respondents in the BRFSS as gay – based on whether they say they are lesbian, homosexual or bisexual – and find that gays have markedly higher mental despair.6 7

Table 2.

Distribution (weighted) of Bad Mental Days and percent with Mental Despair 2020–2025 (N = 1,689,564).

	#Bad MHD	Mental Despair		#Bad MHD	Mental Despair
All	4.76	6.6	Asian	3.47	3.6
Male	4.00	5.7	Hispanic	4.53	6.4
Female	5.48	7.5	Gay	9.64	3.4
Age <25	6.95	7.7	Straight	4.37	6.4
Age 25–34	6.04	7.7	Employees	4.51	5.4
Age 35–44	5.17	7.2	Self-employed	3.94	5.4
Age 45–54	4.63	7.0	Unemployed >1yr	7.70	3.2
Age 55–64	4.18	6.6	Unemployed <1yr	7.26	1.4
Age 65–74	3.05	4.9	Homemaker	4.43	6.1
Age 75–80	2.51	4.0	Student	6.79	5.9
White	4.79	6.6	Retired	2.81	4.4
Black	4.97	7.1	Unable to work	10.45	20.7
Native	6.47	11.2	Native age<25	7.33	10.4

In Table 3 we focus on the young and examine changes in their mental health and life satisfaction over time and by labor market status. The first two columns report the distribution of young people aged under-25 across the various labor force status categories in 1993–2015 and then in 2016–2024. They show that working and being a student account for over four-fifths of youngsters. In the period 1993–2015 they accounted for 54.5 % and 28.9 % respectively and in the subsequent period 51.8 % and 33.7 % respectively. The next two columns show despair rates in the 1990s and 2020s by labor force status. There is a big rise in despair among employees from 3.4 % to 8.1 %, and an approximate doubling among the self-employed (5.2 %–10.0 %).8 The final two columns show mean life satisfaction by labor force status in 2005–2010 then again for 2022–2025 for those aged under-25. It has declined irrespective of status, except in the case of homemakers, where it has remained constant.

Table 3.

Labor Force status distributions and despair, 1993–2024 age < 25.

	Distribution		Despair scores %		Life satisfaction
	1993–2015	2016–2024	1990s	2020s	2005–2010	2022–2025
Employee	50.7	46.8	3.3	7.8	3.33	3.17
Self-employed	3.8	5.0	5.2	10.0	3.34	3.22
Unemp>1year	3.1	2.8	5.9	12.2	3.07	2.98
Unemp<1year	7.0	5.8	5.9	12.2	3.15	2.97
Homemaker	4.1	2.1	4.9	7.8	3.35	3.35
Unable to work	1.6	2.0	15.2	16.9	3.02	2.98
Student	28.9	33.7	2.4	5.3	3.41	3.22
Refused	0.6	1.5	3.4	7.4	3.20	3.19
Work (employee + self-emp)	54.5	51.8	3.4	8.1	3.34	3.18
All			3.5	7.7	3.33	3.17
Total	384,331	238,432	80,577	131,430	84,907	38,278

Fig. 2a, Fig. 2ba and b plot despair by single year of age for 2010–2015 and 2020–2025 respectively for workers and non-workers. The hump-shape among non-workers remains constant across time. Among workers, despair declines with age but what is most striking is the increase in despair among young workers. For example, among 18-year-old workers despair has risen from 2.9 percent to 7.6 percent.

Fig. 2a.

Despair by age, workers and non-workers, 2010–2015.

Fig. 2b.

Despair by age, workers and non-workers, 2020–2025.

Focusing exclusively on workers, Fig. 3 shows the trend in despair across five time periods across the age distribution. The most striking finding is the sharp rise in despair in the last two periods, especially for those below age 50. But the rise in despair is strongest among the youngest and especially in 2020–2025. It turns out that the rise in despair among workers is especially important given their numerical importance for the young since, as shown in Table 3, they account for more than half of the young.9

Fig. 3.

Despair by age for workers, 1993–2025.

Again, focusing exclusively on workers, Fig. 4, tracks changes in despair over time for those aged under 25 and those aged 25 and above. Between 1993 and 2010 the incidence of despair among young workers and older workers was similar. From 2010 onwards, despair rose for both age groups, but much more strongly among the young such that, by the end of the period, 7.5 % of under-25s in paid work were in despair compared to 5.2 % of those workers aged 25 and over.

Fig. 4.

Despair among workers aged <25 and ≥ 25.

Fig. 5 shows how despair has changed by labor market status among the young since 1993. The red dotted line shows the rise in despair among workers presented in Fig. 4. The rise in despair for this group since 2010 is not replicated for other groups. Instead, they show a gentle rise in despair. Whereas in the early 1990s, young people in paid work were less likely to be in despair than other groups, by 2024 the incidence of despair among young workers was similar to that of young students and the young unemployed.

Fig. 5.

Despair among workers and non-workers aged under-25.

This rise in the incidence of despair among young workers from around 2010 has been accompanied by an improvement in the employment rates of the young. As shown in Fig. 6, rates of employment among the young were only slightly lower than those for those aged 25 and above, but this changed in the first decade of the 2000s when young people's employment rates dropped by nearly 18 percentage points, compared to a much shallower decline for those aged 25 and above. But since then, the employment rate among those aged 25 and over has been stable whereas younger people's employment rates have been rising, such that the gap in employment rates by the end of the period narrowed considerably. The literature tends to find that individuals with better mental health are best able to enter employment and hold onto a job, and that very poor mental health acts as a barrier to employment (Frijters et al., 2014). Yet, taken together, Fig. 5, Fig. 6 indicate that young people's employment rates were rising at the same time that their mental health fell.

Fig. 6.

Percent in paid work.

Fig. 7 switches to life satisfaction and shows changes by age and work status over the period 2005–2025. There has been a big drop in life satisfaction, for young workers, whereas for those aged 45 and above, it has remained constant. Life satisfaction for those under age 40 has also dropped for non-workers but, again, little has changed for non-workers aged 40 and above.

Fig. 7.

Life satisfaction for workers and non-workers, 2005–2025.

Table 4 reports on time series changes in despair from 2012 to 2024. Panel a) presents trends by cohort. There has been little recent change in despair among those born before 1965 (column 1). There has been some increase in despair in the decade prior to 2024 for more recent birth cohorts, but by far the biggest change is among the youngest cohort born since 1995 whose despair has almost doubled over the period (from 3.9 percent in 2013 to 7.3 percent in 2024).

Table 4.

Percent in despair over time by cohort and characteristics of those age <25.

a) Cohort
	Pre 1965	1965–1974	1975–1984	1985–1994	1995+	2000+
2013	5.6	6.0	5.6	5.4	3.9
2014	5.6	6.2	6.0	5.3	3.8
2015	5.4	6.5	5.4	5.3	4.2
2016	5.5	6.5	5.6	6.0	4.6
2017	5.6	6.5	6.1	6.1	6.0
2018	5.5	6.7	6.1	6.9	6.9	6.2
2019	5.3	6.4	6.6	7.1	7.2	6.7
2020	5.0	6.6	6.8	6.4	6.6	6.6
2021	5.3	6.8	6.8	7.3	7.8	8.3
2022	5.4	7.2	7.4	8.0	8.8	8.6
2023	5.3	6.9	7.7	8.2	8.0	8.0
2024	4.8	7.2	7.5	8.0	7.3	7.0

b) Despair by race for those age<25 (%)
	White	Black	Asian	Native	Hispanic
2012	5.0	5.7	2.6	6.5	4.9
2013	5.0	4.4	3.1	10.5	4.5
2014	5.1	4.8	1.9	3.3	3.9
2015	5.0	5.3	2.1	6.0	4.5
2016	6.0	4.7	3.2	5.1	4.4
2017	6.4	6.3	3.8	10.8	5.5
2018	7.7	7.1	2.7	7.5	6.7
2019	7.8	6.4	3.9	8.1	7.1
2020	7.6	5.8	3.1	12.2	5.7
2021	9.1	8.1	3.3	12.4	6.6
2022	9.7	9.1	5.0	9.8	7.8
2023	9.0	8.7	3.3	9.4	7.4
2024	7.5	6.6	4.3	8.7	6.0

c) Despair by educational attainment age<25
	Grades 9-11	HSG Graduate	1–3yrs college	≥4 years college
2012	7.9	5.1	4.3	2.5
2013	7.1	4.9	4.3	2.4
2014	6.6	4.6	4.4	2.1
2015	7.1	4.7	4.7	2.6
2016	8.6	5.4	4.7	3.2
2017	9.0	7.2	5.5	2.8
2018	10.2	7.3	6.9	3.8
2019	11.1	7.7	6.9	3.7
2020	9.5	7.3	6.3	3.7
2021	10.8	9.2	7.0	4.4
2022	11.9	9.9	8.2	5.9
2023	11.0	9.4	6.9	4.7
2024	8.9	7.4	7.0	3.9

d) Despair by gender age<25
	Male	Female
2012	4.0	5.9
2013	3.5	6.1
2014	3.9	5.2
2015	4.0	5.5
2016	3.9	6.5
2017	5.4	6.9
2018	5.3	8.9
2019	5.8	8.7
2020	5.1	8.2
2021	6.4	9.7
2022	7.3	10.8
2023	6.8	9.7
2024	5.5	8.4

Panel b) shows trends in despair among the young by race. The rise in despair is particularly marked for whites though it is apparent in all racial groups. Despair rises across the young regardless of educational attainment in panel c). We found that 10.4 % of non-college white workers in 2023 are in despair compared with 5.8 % in 2012. In contrast 6.4 % of white workers with some college, were in despair in 2023 versus 3.4 % in 2012.

Finally, panel d) confirms that, whilst despair rose for both young men and women, the rise was particularly pronounced among young women.

We also examined variance in despair by educational attainment for those of working age from 18 to 65 for the years 2020–2024. The function slopes steeply from left to right for the three lowest education groups. However, the line for those with at least a four-year college degree is much shallower. One possible explanation for these differences by education is that college education proxies for better quality paid work, effectively protecting the young from the distress that may come from engaging in lower quality paid work. Running an equivalent plot for the period 2010–2019 shows a much flatter function for the lowest three education groups with little change for those with a 4-year degree. The difference between 2010-2019 and 2020–2024 may indicate a decline in job quality, or working conditions, for those with lower levels of education, which has become apparent only in the most recent period.10

3.2. Econometric evidence

The changes in the age profile of despair identified in these charts and tables are also illustrated with a series of regression equations that provide ceteris paribus estimates controlling for gender, year and state.

Table 5 reports the results of regressing despair on age, gender, year and state dummies for the five time periods. Sample sizes range between 875,000 and nearly 3 million observations. Part 1 of the table is for all respondents. We find evidence of hump shapes in age maximizing in the first four periods in the age 45–54 age range. In contrast despair declines in age in the most recent period. Part 2 of Table 5 re-runs the analyses but for non-workers only and shows hump shapes in all periods.

Table 5.

Despair regressions over time for all, non-workers, all workers, employees and self-employed.

1) All
	1993–1999	2000–2009	2010–2015	2016–2019	2020–2025
Age 25-34	.0031 (3.47)	−.0001 (0.20)	.0068 (8.99)	.0009 (1.01)	−.0000 (0.02)
Age 35-44	.0106 (12.22)	.0057 (8.76)	.0091 (12.62)	.0010 (1.18)	−.0018 (2.23)
Age 45-54	.0164 (18.02)	.0165 (25.87)	.0209 (30.17)	.0056 (6.67)	−.0056 (7.23)
Age 55-64	.0110 (11.24)	.0098 (15.29)	.0156 (22.97)	.0013 (1.57)	−.0103 (13.67)
Age 65-74	−.0002 (0.28)	−.0107 (16.03)	−.0084 (12.21)	−.0212 (26.08)	−.0313 (42.08)
Age 75-80	.0027 (2.53)	−.0150 (21.74)	−.0167 (23.51)	−.0322 (37.84)	−.0415 (53.86)
Female	.0153 (33.95)	.0162 (61.02)	.0141 (51.00)	.0154 (44.11)	.0181 (55.75)
Adjusted R2	.0061	.0062	.0069	.0070	.0070
N	873,937	2,978,477	2,758,188	1,765,927	2,139,078

2) Non-workers
	1993–1999	2000–2009	2010–2015	2016–2019.	2020–2025
Age 25-34	.0270 (14.10)	.0315 (23.64)	.0454 (32.35)	.0048 (26.78)	.0431 (26.82)
Age 35-44	.0616 (32.05)	.0682 (53.02)	.0769 (56.33)	.0813 (46.49)	.0747 (47.88)
Age 45-54	.0822 (41.51)	.1071 (86.84)	.1141 (91.18)	.1047 (66.18)	.0850 (58.16)
Age 55-64	.0290 (16.39)	.0383 (33.49)	.0538 (47.30)	.0486 (34.92)	.0379 (29.72)
Age 65-74	−.0005 (0.28)	−.0110 (9.90)	−.0062 (5.66)	−.0145 (10.94	−.0213 (18.04)
Age 75-80	.0021 (1.23)	−.0173 (15.50)	−.0176 (15.92)	−.2933 (21.87)	−.0356 (29.81)
Female	−.0015 (1.65)	−.0004 (0.78)	.0019 (4.12)	.0035 (6.14)	.0071 (13.62)
Adjusted R2	.0189	.0287	.0303	.0303	.0262
N	324,038	1,296,296	1,382,171	867,671	1,020,788

3) Workers
	1993–1999	2000–2009	2010–2015	2016–2019.	2020–2025
Age 25-34	−.0013 (1.37)	−.0076 (10.62)	−.0054 (6.86)	−.0126 (13.18)	−.0128 (13.81)
Age 35-44	.0017 (1.81)	−.0081 (11.72)	−.0097 (12.64)	−.0204 (21.76)	−.0218 (24.17)
Age 45-54	.0015 (1.51)	−.0078 (11.31)	−.0106 (14.22)	−.0272 (29.84)	−.0335 (37.52)
Age 55-64	−.0049 (4.32)	−.0119 (16.63)	−.0131 (17.44)	−.0323 (35.56)	−.0423 (47.09)
Age 65-74	−.0116 (6.59)	−.0218 (23.85)	−.0216 (24.73)	−.0394 (37.62)	−.0507 (49.16)
Age 75-80	−.0091 (2.56)	−.0217 (14.75)	−.0228 (17.41)	−.0427 (26.56)	−.0523 (33.44)
Female	.0153 (30.49)	.0160 (14.75)	.0128 (52.329)	.0131 (32.06)	.0183 (44.39)
Adjusted R2	.0042	.0036	.0029	.0054	.0075
N	545,637	1,673,779	1,358,164	880,295	1,078,050

4) Employees
	1993–1999	2000–2009	2010–2015	2016–2019.	2020–2025
Age 25-34	−.0011 (1.13)	−.0075 (10.15)	−.0054 (6.55)	−.0135 (13.53)	−.0129 (13.25)
Age 35-44	.0021 (2.21)	−.0077 (10.80)	−.0097 (12.14)	−.0212 (21.49)	−.0222 (23.30)
Age 45-54	.0019 (1.92)	−.0074 (10.35)	−.0106 (13.55)	−.0278 (28.85)	−.0339 (35.67)
Age 55-64	−.0045 (3.71)	−.0114 (15.07)	−.0128 (16.26)	−.0323 (33.60)	−.0422 (43.98)
Age 65-74	−.0121 (5.88)	−.0222 (21.58)	−.0221 (22.86)	−.0400 (34.09)	−.0503 (43.18)
Age 75-80	−.0135 (2.91)	−.0244 (13.02)	−.0257 (15.48)	−.0428 (20.38)	−.0551 (26.18)
Female	.0167 (31.00)	.0169 (53.12)	.0142 (41.62)	.0144 (31.78)	.0200 ()
Adjusted R2	.0046	.0037	.0054	.0073
N	471,008	1,409,557	1,131,052	723,330	892,206

5) Self-employed workers
	1993–1999	2000–2009	2010–2015	2016–2019.	2020–2025
Age 25-34	−.0101 (2.53)	−.0127 (4.78)	−.0091 (3.16)	−.0062 (1.94)	−.0130 (4.35)
Age 35-44	−.0110 (2.87)	−.0176 (6.94)	−.0149 (5.43)	−.0183 (5.89)	−.0224 (7.91)
Age 45-54	−.0120 (3.07)	−.0182 (7.28)	−.0175 (6.53)	−.0278 (9.14)	−.0359 (12.80)
Age 55-64	−.0187 (4.61)	−.0239 (9.44)	−.0219 (8.25)	−.0359 (11.96)	−.0477 (17.19)
Age 65-74	−.0244 (5.33)	−.0327 (12.13)	−.0299 (10.90)	−.0429 (13.92)	−.0582 (20.49)
Age 75-80	−.0173 (2.69)	−.0308 (9.75)	−.0292 (9.57)	−.0476 (13.65)	−.0570 (17.87)
Female	.0076 (5.46)	.0131 (17.74)	.0081 (10.57)	.0089 (9.10)	.0127 (12.85)
Adjusted R2	.0033	.0041	.0035	.0065	.0090
N	74,629	264,222	227,112	156,995	185,884

Includes year and state dummies, 18–24 excluded. Samples include Guam, Puerto Rico and USV.

Part 3 reports analyses for all workers. Parts 4 and 5 then split this group out into employees and the self-employed respectively. Here the picture is entirely different. Despair for workers declines in age, but this pattern becomes more pronounced over time. The picture is broadly similar for employees and the self-employed, though the rise in despair is greater among employees.11

Table 6 now separates out non-workers into the four main categories – the unemployed, students, homemakers and those who are unable to work. We exclude the retired as there are so few young people in this category. For the unemployed, students and those unable to work, despair rises in age until middle-age and then declines. But for homemakers it declines in age. These patterns are apparent across all periods. For the former three groups everywhere, there is a significant and positive female coefficient. In contrast it is significantly negative for homemakers in the four periods since 2000.

Table 6.

Despair regressions over time for types of non-worker.

1) Unemployed
	1993–1999	2000–2009	2010–2015	2016–2019	2020–2025
Age 25-34	.0259 (5.44)	.0256 (7.44)	.0333 (10.00)	.0259 (5.44)	.0228 (5.77)
Age 35-44	.0499 (10.24)	.0501 (15.17)	.0506 (15.33)	.0498 (10.24)	.0415 (10.47)
Age 45-54	.0490 (10.66)	.0610 (19.07)	.0604 (19.67)	.0490 (10.66)	.0385 (9.76)
Age 55-64	.0189 (4.23)	.0378 (11.12)	.0373 (12.17)	.0189 (4.23)	.0156 (4.04)
Age 65-74	−.0209 (3.68)	−.0016 (0.31)	−.0053 (1.29)	−.0209 (3.68)	−.0257 (5.25)
Age 75-80	−.0498 (6.24)	−.0283 (3.81)	−.0286 (4.72)	−.0498 (6.24)	−.0482 (6.53)
Female	.0197 (7.93)	.0186 (10.03)	.0189 (11.30)	.0197 (7.93)	.0187 (8.73)
Adjusted R2	.0103	.0086	.0084	.0103	.0096
N	70,457	126,773	146,103	70,457	95,962

2) Students
	1993–1999	2000–2009	2010–2015	2016–2019	2020–2025
Age 25-34	.0180 (6.60)	.0218 (10.75)	.0265 (13.03)	.0174 (6.49)	.0153 (5.71)
Age 35-44	.0375 (8.92)	.0372 (12.90)	.0472 (15.34)	.0329 (7.30)	.0487 (11.12)
Age 45-54	.0246 (3.43)	.0502 (13.04)	.0485 (12.04)	.0376 (6.24)	.0423 (6.98)
Age 55-64	.0294 (1.95)	.0274 (4.11)	.0300 (4.77)	.0392 (4.31)	.0228 (2.34)
Age 65-74	.0142 (0.83)	.0136 (1.41)	.0062 (0.64)	−.0205 (1.62)	−.0152 (1.24)
Age 75-80	.0030 (0.13)	−.0173 (1.49)	.0299 (2.31)	−.0270 (1.65)	−.0071 (0.49)
Female	.0185 (8.27)	.0197 (11.35)	.0177 (10.67)	.0210 (10.09)	.0314 (15.30)
Adjusted R2	.0079	.0115	.0104	.0061	.0097
N	28,592	64,476	64,755	46,783	42,422

3) Homemakers
	1993–1999	2000–2009	2010–2015	2016–20192	020–2025
Age 25-34	−.0074 (2.04)	−.0131 (5.27)	−.0112 (3.17)	−.0104 (1.98)	−.0071 (1.30)
Age 35-44	−.0040 (1.08)	−.0155 (6.27)	−.0178 (5.10)	−.0211 (4.05)	−.0153 (2.85)
Age 45-54	.0057 (1.49)	−.0064 (2.56)	−.0091 (2.58)	−.0186 (3.54)	−.0181 (3.32)
Age 55-64	.0003 (0.09)	−.0088 (3.43)	−.0108 (3.07)	−.0198 (3.76)	−.0182 (3.34)
Age 65-74	−.0097 (2.35)	−.0206 (7.93)	−.0232 (6.53)	−.0301 (5.62)	−.0309 (5.51)
Age 75-80	−.0044 (0.97)	−.0257 (9.98)	−.0317 (9.12)	−.0449 (8.59)	−.0441 (8.07)
Female	.0022 (0.28)	−.0145 (3.76)	−.0237 (6.32)	−.0176 (3.84)	−.0230 (5.62)
Adjusted R2	.0076	.0056	.0059	.0071	.0054
N	69893	236,352	183,356	91,633	85,999

4) Unable to work
	1993–1999	2000–2009	2010–2015	2016–2019	2020–2025
Age 25-34	.0505 (3.10)	.0926 (10.23)	.0681 (7.04)	.0658 (5.76)	.0462 (4.74)
Age 35-44	.0991 (6.49)	.1200 (14.36)	.0958 (10.58)	.0927 (8.63)	.0858 (9.49)
Age 45-54	.0928 (6.20)	.1166 (14.36)	.0769 (8.82)	.0775 (7.49)	.0737 (8.50)
Age 55-64	.0292 (1.95)	.0436 (5.39)	.0212 (2.45)	.0192 (1.88)	.0165 (1.95)
Age 65-74	−.0147 (0.90)	−.0205 (2.43)	−.0345 (3.87)	−.0175 (1.67)	−.0193 (2.21)
Age 75-80	−.0186 (0.99)	−.0574 (6.45)	−.0886 (9.58)	−.0773 (6.99)	−.0598 (6.39)
Female	.0273 (3.10)	.0288 (13.51)	.0287 (14.59)	.0309 (12.95)	.0312 (13.37).
Adjusted R2	.0220	.0221	.0178	.0150	.0138
N	28883	172,850	197,644	127,395	126,809

Includes year and state dummies, 18–24 excluded. Samples include Guam, Puerto Rico and USVI.

Recall, from Fig. 6 the proportion of the young in work has changed markedly over the period. Below we report the weighted estimates in the BRFSS data of the changes over time in the percentage employed. Youth employment rates fell sharply after the Great Recession and more rapidly than for those aged 25 and over. But they picked up more rapidly from around 2010, so the rising despair rates of young workers become relatively more important as their share of the young begins to rise. There is no change over these years in the employment rates of older workers, which stayed steady at around 58 %.

% working	Age<25	Age≥25
1993	60 %	63 %
2000	63 %	64 %
2008	51 %	61 %
2009	45 %	59 %
2010	45 %	58 %
2011	46 %	57 %
2012	47 %	57 %
2016	51 %	58 %
2018	53 %	58 %
2020	50 %	57 %
2022	55 %	58 %
2024	54 %	57 %

Table 7 reports life satisfaction equations for workers and non-workers for the periods 2005–2018 and 2022–2025. The evidence is broadly consistent with the evidence on despair. In both time periods life satisfaction of workers rises in age whereas there are U-shapes in age for non-workers.

Table 7.

Life satisfaction.

	Workers		Non-workers
	2005–2018	2022–2025	2005–2018	2022–2025
Age 25–34	.0703 (23.45)	.1162 (24.93)	−.0495 (11.90)	−.0173 (2.25)
Age 35–44	.0719 (25.02)	.2086 (46.33)	−.1264 (31.74)	−.0332 (4.49)
Age 45–54	.0663 (23.44)	.2512 (56.15)	−.2430 (64.49)	−.0580 (8.34)
Age 55–64	.1025 (35.59)	.2989 (66.67)	−.0292 (8.31)	.0799 (13.22)
Age 65–74	.1514 (44.67)	.3396 (66.84)	.1034 (30.09)	.2488 (44.18)
Age 75–80	.1458 (29.19)	.3578 (48.20)	.0710 (20.62)	.2555 (45.20)
Female	−.0052 (4.94)	.0093 (4.59)	.0312 (23.39)	.0357 (15.68)
Adjusted R2	.0047	.0258	.0292	.0334
N	1,267,054	336,440	1,133,210	335,722

Equations include year and state dummies. T-statistics in parentheses. BRFSS, 2005–2025.

There is prior evidence that the relation between age and ill-being is different between workers and non-workers. Blanchflower and Bryson (2022) showed that what matters in explaining pain through to age 65 is also whether one is working or not. Pain rose with age. “In the United States and elsewhere pain rises with age among workers but is hump shaped among non-workers, maximizing around age 50 and declining thereafter.” This was found using Gallup World Poll and US Daily Tracker data (2008–2017). In Appendix Fig. 1 we report results of regressing pain on, single year of age, gender, year, and country and plucking the coefficients for OECD countries, 2010–2025 by workers (n = 243,788) and non-workers (n = 182,993). Thus, the age pattern in pain among non-workers reflects that for despair, whereas among workers pain rises in age while despair falls in age.

In Appendix Fig. 2 we report on the relation between sadness and age by work status.12 Again, there are marked differences between workers and non-workers. The age-pattern in sadness is hump-shaped for non-workers but is flat for workers.

3.3. Mental distress and age by work status, other surveys

Data is also available on serious psychological distress from the National Survey on Drug Use and Health (NSDUH); which is an annual survey of the U.S. population, including individuals 12 years of age and older. It oversamples adolescents and young adults. Adult respondents (18 years of age and older) completed the Kessler-6 Distress Scale, a reliable validated scale that asks respondents how frequently they experienced symptoms of psychological distress during the past 30 days. The six symptoms were: feeling nervous, feeling hopeless, feeling restless or fidgety, feeling so sad or depressed that nothing could cheer you up, feeling that everything was an effort, and feeling down on yourself, no good, or worthless.

Response choices were coded as 4 (all of the time), 3 (most of the time), 2 (some of the time), 1 (little of the time), and 0 (none of the time). The possible range of scores was 0–24. Scores of 13 and over were coded by the survey administrators as indicative of serious psychological distress. Following Twenge et al. (2019) we relied on this dichotomous variable in our analyses. Psychological distress declines in age, in these data since 2008. The authors showed a rise in psychological distress especially for ages 18–19 and 20–21.13 We extend these estimates and confirm they rose especially for workers.

Table 8 shows the rise in the Kessler-6 psychological distress score and extends the Twenge et al. (2019) results through to 2023 (n = 639,450). We examine the percentage above the cutoff of 12 as being in distress. Distress rises particularly among the young from around 2014. By 2023 around one-in-five aged 18–23 reported being in distress. This was particularly apparent among female workers. Roughly one-quarter of female workers under age 24 were in distress by 2023 versus around 14 % of male workers. There has been a decline since the COVID peak in 2021 for young workers.

Table 8.

Serious Psychological Distress (Kaiser score >12) (weighted), NSDUH n = 639,450.

Time/age	18–20	21–23	24–25	26–29	30–34	35–49	50–64	65+
2008	9	7	6	6	5	5	3	3
2009	8	8	7	5	6	5	4	2
2010	9	8	6	6	5	5	4	2
2011	9	7	7	7	5	5	4	2
2012	9	8	8	6	6	5	4	3
2013	10	8	6	7	6	5	4	3
2014	10	9	8	5	5	5	4	3
2015	12	10	10	7	5	5	4	2
2016	14	10	9	7	6	6	5	2
2017	15	13	11	9	7	6	4	2
2018	16	14	11	10	7	5	4	2
2019	17	16	14	12	8	6	4	3
2020	19	17	15	11	9	7	5	2
2021	22	19	16	13	10	7	4	2
2022	21	19	17	14	11	7	5	2
2023	19	20	15	16	11	7	4	2
N	89,593	89,978	61,136	52,970	64,704	156,283	70,990	53,796
Male workers
2021	15	13	4	11	8	7	3	2
2022	17	14	6	16	12	8	5	3
2023	15	13	5	12	12	8	4	1
Female workers
2021	31	23	7	19	15	10	6	3
2022	25	24	6	17	14	11	5	3
2023	23	25	7	17	19	10	7	3

3.4. Possible explanations for the poor well-being of young workers

The analyses presented in Section 3.3 raise a big question: why has the mental health of young workers declined in the BRFSS so much since 2015? The trend began well before COVID, so the pandemic cannot fully account for the change.

One possibility, noted at the outset, is that the relative wage of youth jobs has fallen. But this doesn't seem to be the case. If we look at median, usual hourly earnings in current dollars reported by the Bureau of Labor Statistics for ages 16–24, taken from the Current Population Survey, the ratio compared to workers age 25 and older is little changed. This is reported in Fig. 8a. Since around 1985 the ratio has increased over time. We found it rose from 60 % in 1985 to 75 % in 2024.

Fig. 8a.

Annual ratio usual hourly earnings age 16–24/25+.

In Fig. 8b we report quarterly estimates of current weekly earnings since 2000. We found the youth/adult earnings ratio rose from 56.6 % in 2015Q1 to 60.0 % in Q22025. The youth wage was $487 in 2015 compared with $860 for the 25+ group. In 2025Q2 weekly wages were $758 and $1264 respectively. Plus, the real weekly wage is up, over the last decade for employees. For example, real weekly wages of private sector employees have risen (in 1982–1984 dollars). They average $385.52 in the first five months of 2025, up 2.4 % since 2019, but down 2 % since 2021 after the wage burst in 202.

Fig. 8b.

Quarterly ratio usual weekly earnings age 16–24/25+.

A second possibility is that the rise of the gig economy has severely impacted younger workers. As noted earlier, whilst beneficial for some, gig working can create uncertainty and insecurity for those who resort to it because they have no clear path to permanent employment and career progression. Although Katz and Krueger (2019) suggested only a modest rise in the number of gig workers between 2000 and 2017 and argued that workers had a hard time reporting their work status in standard surveys, a recent survey from the Federal Reserve – the 2024 Survey of Household Economics and Decision making (SHED) - suggests that gig economy work, very broadly defined, is especially important for the young ages 18–29, although the difference compared to older adults is small.14

Another third possible explanation is the rise in under-employment among the young. The underemployed are those who have a job offering fewer hours than the worker would like. The underemployed have low levels of job satisfaction and overall happiness (Bell & Blanchflower, 2019). Since the Great Recession there has been a marked increase in underemployment among the young. Underemployment is generally measured in the US using data from the Current Population Survey, where individuals are asked if they are part-time for economic reasons (PTFER). Bell and Blanchflower (2021) used this to derive an under-employment measure which is simply PTFER/employment.

The Merged Outgoing Rotation Group (MORG) files of the Current Population Survey for 2009–2024 allow us to identify workers who are PTFER. In Table 9 we estimate the probability of underemployment with a set of age controls as well as gender, year, state and union membership. We find evidence that the probability of being underemployed declines in age. Below, for workers we see that the underemployment rate for the young declined steadily from a peak of 2.14 % through 2020 and jumped sharply during COVID before declining. However, it continues to be relatively higher than the rate for older workers.

	% underemployed age 18-24	Ratio vs over 25 rate
2009	2.14	99 %
2010	1.71	97 %
2011	1.74	117 %
2012	1.80	138 %
2013	1.38	108 %
2014	1.29	119 %
2015	1.16	116 %
2016	1.30	137 %
2017	1.09	115 %
2018	1.03	115 %
2019	1.14	131 %
2020	2.40	91 %
2021	0.96	87 %
2022	1.18	126 %
2023	1.18	133 %
2024	1.03	112 %

Table 9.

Underemployment – modeling the proportion part-time for economic reasons among workers. MORG files of the CPS.

	2009–2024	2009–2019	2020–2024	2020–2024
Age ∗100				−.0019 (2.12)
Age 25–34	−.0023 (9.84)	−.0024 (8.77)	−.0021 (4.52)
Age 35–44	−.0031 (13.25)	−.0034 (12.38)	−.0025 (5.15)
Age 45–54	−.0032 (13.61)	−.0037 (13.66)	−.0018 (3.63)
Age 55–64	−.0039 (15.70)	−.0044 (15.28)	−.0026 (5.13)
Age 65–74	−.0033 (9.02)	−.0041 (9.36)	−.0014 (2.12)
Age 75–80	.0022 (3.12)	−.0033 (3.82)	.0002 (0.17)
Female	−.0023 (17.60)	−.0026 (17.28)	−.0015 (5.44)	−.0014 (5.42)
Union	−.0006 (2.79)	−.0006 (2.40)	−.0006 (1.39)	−.0008 (1.67)
Constant	.0216	.0221	.0246	.0235
Adjusted R2	.0020	.0016	.0033	.0033
N	2,375,802	1,750,509	625,293	625,293

Equations are for workers only and include year and state dummies. Excluded 18-24.

A fourth possibility is that there has been a reduction in job satisfaction among the young, potentially reflecting a decline in job quality or else an increase in expectations about what a job might offer. In the US General Social Surveys 1972–2024 workers have reported how satisfied they are with their jobs.15 Averaged across the 2021, 2022 and 2024 surveys, for example, this variable rises linearly in age, with the young being the least happy.16 Job satisfaction also rose linearly in the earlier period 2000–2018 when the young also had the lowest levels of satisfaction and their level declined from 3.12 to 3.04. The fall in job satisfaction in this table is larger at younger ages.

	2021–2024	2000–2018
Age <25	3.04	3.12
25-34	3.17	3.27
35-44	3.27	3.34
45-54	3.32	3.36
55-64	3.42	3.44
65+	3.60	3.63
N	5,702	13,789

If, as suggested in the introduction, there may have been a decline in job quality among the young in particular, we might expect the increase in despair to be most apparent among those young people without college education. This is apparent in Fig. 9, which shows trends in despair for the college-educated and non-college-educated under-25s in employment in BRFSS. It is apparent that despair has always been a little higher among less-educated workers, but the gap has increased since around 2014.

Fig. 9.

Despair among age<25 workers.

Other evidence is consistent with the declining well-being of young workers. An annual survey of more than 1.5 million individuals at over 2,500 organizations in the U.S. found that workplace well-being from 2019 through 2023 spiked at the start of the pandemic in 2020 and has since declined as workers have returned to offices and lost some of the flexibility that had provided work-life balance.17 But well-being declined most among the young. Smith et al. (2024) say “we would generally find a declining score in well-being scores with advancing age groups. However, between 2020 and 2023, this trend reversed, indicating an increase in well-being scores with age”. By 2023 there was “a near linear relationship between age and well-being”.

In addition, the Pew Research Center found that younger workers in 2024 are among the least satisfied with their jobs. Lin et al. (2024) found that younger workers in 2024 had especially low levels of job satisfaction. The percentage of workers who were satisfied with their jobs was as follows.

	Extremely/very	Somewhat	Not too/not at all
Ages 18–29	43	40	17
Age 30–49	48	41	11
Age 50–64	56	34	10
Age 65+	67	27	6

A report by the Conference Board (2025) confirms this finding and suggested that the job satisfaction gap between younger and older workers continued to widen. It found a 15-point gap in job satisfaction between the oldest and youngest generations. Only 57.4 % of US workers under age 25 reported being satisfied with their jobs compared to 72.4 % of those aged 55 and older. In addition, Cangrade (2024) investigated happiness at work and found the youngest were once again, the most unhappy. The proportions who said they were unhappy were as follows by generation - Baby Boomers (born 1955–1964) = 9 %; Gen X (born 1965–1980) = 13 %; Millennials (born 1981–1996) = 13 % and Gen Z (born 1997–2012) = 26 %.

Perhaps for these reasons, we also find that young workers are increasingly favourable to trade unions. For example, in a recent report the EPI argued … “over 40 % of nonunion workers 30 and under are outright supporters of unionization, whereas only 32 % of older nonunion workers hold the same view; union opposition is much stronger among older workers than among younger ones". 18 Data from Gallup for 2024 (page 8) shows that 77 % of respondents age 34 and lower say they approve of labor unions—higher than any other age group—compared to 70 percent of respondents age 35–54 and 66 % over age 55.19 Aurelia Glass (2024) argues that

"Young workers today feel as though they’ve been left behind by high housing costs, low wage growth, and lower job quality, especially for workers without college degrees. Unions enable workers to advocate collectively for better wages and working conditions, giving young workers a strategy to support themselves. Young workers are finding that union membership helps them achieve better financial stability, with workers under the age of 34 earning 11.3 percent more from union membership."

Twenge et al. (2017) argues there has been a reconsideration of work among the young. Especially post-COVID, she suggests, the work ethic has plummeted. She examined data from the nationally representative Monitoring the Future Survey which has asked U.S. 12th graders, most of whom are 18 years old, about their work attitudes since 1976. She notes that the number of 18-year-olds who said they wanted to do their best in their job “even if this sometimes means working overtime” suddenly plummeted in 2021 and 2022. In early 2020, 54 % of 18-year-olds said they were willing to work overtime. By 2022, it was 36 %. That's a (relative) drop of 33 % in just two years. It's also an all-time low in the 46-year history of the survey. Twenge suggests five reasons for this.

• 1.Pandemic burnout
• 2.Pandemic reminded us that life is more than work.
• 3.The job market was stronger in 2021-22 and employees could favor work life balance.
• 4.TikTok made quiet quitting viral.
• 5.Gen-Z is pessimistic about a rigged system.

This reorientation to work is another potential explanation for the growth in despair among the young. It does not require a change in the nature of work itself. However, it is not incompatible with the alternative possibility, namely that young people are reacting to declining job quality (Graeber, 2019).

Our analyses indicate that the mental health of young women is worse than men - there is about a 3-percentage point difference in the levels of despair – but the trend in despair has been steeper among young women. A referee has suggested to us that a plausible explanation why female despair rates are higher than males despite more women entering the labor force, is that they typically still have the larger burden of childcare and housework than that of men. This was particularly noticeable during COVID, when even women's labor force participation suffered because they disproportionately bore the burden of children having to do home learning rather than attend school. The data below are taken from the BLS and are seasonally adjusted estimates of the labor force participation rate of 16–19 and 20–24 age groups for men and women. Male rates have declined for both groups as they have for ages 16–19 girls but have remained broadly constant for women ages 20–24 since 2022. For men age 20–24 LFPR rates are lower in 2025 than in 2012.

LFPR	January 2012	January 2020	January 2022	August 2025
Women age 16-19	34.4	37.0	37.3	35.1
Women ages 20-24	66.9	70.8	68.2	68.1
Men age 1619	32.7	35.6	36.1	34.4
Men age 20-24	75.5	74.9	73.7	72.2

4. Conclusions

In this paper we have confirmed that the mental health of the young in the United States has worsened rapidly over the last decade, as reported in multiple datasets. The deterioration in mental health is particularly acute among young workers and especially young female workers. We have previously shown that there has been a shift in the age profile of poor mental health such that the hump-shape in age has been replaced by a monotonic decline in age, and that this is due to a worsening in the mental health of young people, both in absolute and relative terms. But this is the first paper to show that this change is primarily due to a change in the mental health of young workers. They have experienced a rise in despair that happened pre-COVID but has continued during the COVID period. This is especially the case for the least educated. Most had their schooling during COVID lockdown where their social interactions were limited. Lack of social capital may be a contributory factor (Helliwell & Putnam, 2004; Putnam, 2000).

At the same time that despair has been rising among young workers, the employment rate of the young has risen so that they account for a growing proportion of all young Americans. We conclude that the reason that mental despair now declines in age is driven primarily by the recent decline in the mental health of workers, and especially employees, under the age of forty and especially those under twenty-five.

It does not appear that the declining mental health of young workers is driven by a decline in the youth wage compared to the wage of older workers; this ratio has increased based on the figures we present although, as noted earlier, there is debate over this issue.

Nevertheless, as Feiveson (2024) has noted, the relative prices of housing and childcare have risen, making it increasingly tough for young people and young families to make ends meet and get on the housing ladder. Student debt is high and expensive. The health of young adults has also deteriorated, as seen in increases in social isolation and obesity. Suicide rates of the young are rising. Moreover, Jean Twenge provides evidence that the work ethic itself among the young has plummeted. Some have even suggested the young are unhappy having BS jobs. There is a good deal of supporting evidence from a variety of surveys including from Pew, the Conference Board and Johns Hopkins on the parlous state of young worker well-being in the USA that we documented here.

It is possible that we are observing the consequences of past well-being shocks. We should note that 10.1 % of workers aged 20 in 2023 said they were in despair. They were aged 17 when COVID lockdowns were implemented in 2020. They were 10 years old in 2013 as the smartphone and the internet exploded. In addition, of course, they were in high school ages 14–18 in 2017–2021. We know from the Youth Risk Behavior Survey that the well-being of high school students deteriorated sharply around that time.20

The particularly sharp rise in despair among young, less educated workers might suggest that the new ‘worker problem’ could be linked to a decline in job quality, since there is a strong correlation between getting into good jobs and educational attainment. As noted in the introduction, the recent literature on job quality does indicate that employers are exploiting new technologies to increase labor intensification and tighten their monitoring of worker effort, especially in jobs where young people are concentrated such as gig working, fast food and retail. Although this is speculation on our part, it is consistent with the trends we find are cohort effects, rather than pure age effects.

An important limitation of our analysis is that we take no account of potential differential selection into employment. If we take educational attainment, for instance. At a point in time, we might expect those with lower educational attainment to have a lower probability of employment than more highly educated individuals. We might also anticipate a positive correlation between poor educational attainment and mental despair. This means that, conditional on being in paid work, the descriptive ranking of despair by educational attainment implies a downward bias in the despair expressed by the less educated. It might also be that there is differential selection into employment by educational attainment over time, and that this might affect the trends in despair we see among workers by educational attainment level. These issues are deserving of attention in future work.

CRediT authorship contribution statement

David G. Blanchflower: Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation. Alex Bryson: Writing – review & editing, Validation, Software, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Ethical statement

Ethical approval is not applicable to this manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data publicly available.