For decades, happiness research showed a U-shape across a lifetime — high in youth, bottoming out at around 50, then rising again after 70. A new study across 44 countries finds that curve has flipped: young adults are now the least happy group, and despair declines steadily with age

The old happiness curve was easy to remember. Wellbeing began relatively high in youth, slid towards a trough in middle age, then recovered. It appeared often enough across countries that researchers came to treat the U-shape as one of the sturdier patterns in social science.

A 2025 paper argues that the contemporary picture looks very different. This is not clinical advice. I am reading the research as a writer and asking what it does, and does not, establish.

The finding is worth taking seriously, but it should not be read as the final word. The paper’s 44-country result concerns self-reported mental ill-being and distress, not happiness in every possible sense. Its longer evidence of the curve changing over time comes from the United States and United Kingdom.

That qualification does not make the result small. It makes the actual result clearer: younger adults in these datasets are not merely missing an advantage once associated with youth. Their reported ill-being has risen enough to alter the age profile itself.

The curve was always more contested than the shorthand suggested

The modern case for the U-shape is usually traced to a 2008 paper by David Blanchflower and Andrew Oswald. Using large samples from the US and Europe, and further evidence from dozens of countries, they found that happiness tended to reach its low point in middle age after other circumstances were held constant.

Hundreds of later estimates found something similar. Blanchflower eventually reported U-shaped wellbeing across 145 countries, with a typical trough near 50. The mirror image also appeared: worry, stress and other forms of ill-being tended to rise towards middle age and fall afterwards.

But the neat curve hid a methodological argument. Results can change depending on whether researchers control for employment, marriage, income and health, all of which change with age. Cross-sectional data also struggle to separate ageing from period effects and differences between generations. A 2021 analysis of 81 countries, for example, identified several shapes rather than one universal U.

I previously wrote about the familiar midlife dip. Reading this newer work, I would now put more weight on the date and measure behind any such curve. A population pattern can be real for one era without becoming a law of human development.

What the new paper actually combined

The paper, published in PLOS One, was written by Blanchflower of Dartmouth College, Alex Bryson of University College London and Xiaowei Xu of the Institute for Fiscal Studies. It joins three distinct sources rather than presenting one worldwide longitudinal survey.

For the US, the team used the Centers for Disease Control and Prevention’s Behavioral Risk Factor Surveillance System, a telephone survey that now conducts more than 400,000 adult interviews a year. The analysis spans 1993 to early 2024. Its starkest measure, labelled “despair” by the authors, meant reporting that mental health was not good on all 30 of the previous 30 days.

For Britain, the researchers drew on Understanding Society, the UK household longitudinal study, from 2009 to 2023, along with a separate national anxiety measure. They created a despair threshold from the 12-item General Health Questionnaire.

The global section used the Global Minds Project, an anonymous online assessment available to adults. The researchers retained 44 countries with at least 10,000 responses each, giving roughly 1.7 million records from people aged 18 to 74 between 2020 and 2025. They examined an overall Mental Health Quotient and measures covering distress, fear and anxiety, sadness or hopelessness, and suicidal thoughts.

These instruments are related, but they are not identical. “Despair” is an analytical label with different operational definitions in the US and UK sections. It is not a diagnosis assigned to each respondent.

The reversal came from younger people getting worse

The US time series shows the direction of change most plainly. The share of people under 25 meeting the paper’s despair definition rose from 2.9 per cent in 1993 to 8 per cent in 2023. Looking at the more recent period, the increase was particularly steep among young women. Meanwhile, the age profile for people in their late forties and beyond changed far less.

In the earlier US period, ill-being formed the expected hump, with higher levels around middle age. From 2019 onwards it generally declined with age. The curve did not flip because older people suddenly became carefree. It flipped mainly because younger groups deteriorated relative to them.

The British data tell a similar story. Among people under 25, the share of young men above the paper’s despair threshold rose from 2.3 per cent in 2009 to 6.4 per cent in 2021. For young women it rose from 4.4 to 12.7 per cent. The old hump-shaped pattern disappeared in the later period, while anxiety also shifted towards a decline with age.

Across the Global Minds sample, 13.4 per cent of under-25s fell into the assessment’s most distressed category, compared with 5.6 per cent of respondents overall. The proportion declined more or less steadily through each older age band in all 44 countries. Regression models produced the same broad gradient after accounting for country, gender, education, labour-force status, survey year and even the timing of the response.

Why this is not quite a global happiness inversion

The title-sized version is that young people are now the least happy everywhere. The data require more restraint.

Global Minds is a large online dataset, but its assessment is open to anyone with internet access who chooses to complete it. It is not the same as drawing a probability sample of every country’s population. A sample of 1.7 million can estimate a pattern very precisely while still carrying selection effects.

The global window also begins in 2020. It shows that ill-being declined with age during 2020 to 2025, but it cannot by itself show the earlier curve flipping within each of those 44 countries. The historical reversal is demonstrated most directly by the longer US and UK series.

Other international evidence depends on the outcome used. The World Happiness Report 2024, which measured people’s evaluations of their lives across 143 countries, found young people were the least happy group in only seven. Older people were the least happy group in 71. Yet it also found a strong youth decline in North America and parts of Western Europe, with the young least happy in the US and Canada.

That is not a clean contradiction. Feeling intense distress, evaluating one’s life poorly and answering a happiness question are different outcomes. The new paper’s force lies in finding a similar age gradient across several measures of ill-being, not in proving that every country now shares one happiness curve.

The study maps the change better than it explains it

The authors consider several possible causes: weaker labour-market prospects after the financial crisis, housing pressure, stretched services, the pandemic, smartphones and social media. They note that the youth decline began before COVID, although the pandemic appears to have intensified it.

None of the three datasets used here can apportion responsibility among those explanations. The global analysis is observational, and the US and UK trends do not contain an experiment capable of isolating one cause. The paper discusses external studies on smartphones at length, but its own central models establish an age pattern, not a mechanism.

There is a broader structural point worth keeping. If the curve changed because younger people became more distressed, framing older adults as unusually resilient misses the event. It also turns a cohort facing different economic, social and technological conditions into a morality tale about attitude.

The study likewise does not tell us what happens as today’s young respondents grow older. Repeat cross-sections compare age groups at different moments. They cannot show whether this cohort will improve with age, carry its current disadvantage forward, or follow another path entirely.

A broken curve is a warning, not a verdict

The U-shape offered an oddly comforting story about middle age. The low point was normal, and the upward turn was waiting. The new evidence removes that reassurance for young adults, but it should not replace it with a prediction that an unhappy youth guarantees an unhappy life.

What it does offer is a useful correction to where concern is directed. A framework built around the midlife crisis can fail to notice a faster deterioration earlier in adulthood. Institutions designed around yesterday’s age profile may continue looking for strain in the wrong place.

If this subject feels personal rather than statistical, and distress is persistent or severe, a qualified health professional or an appropriate local support service is more useful than any article.

The curve is not a natural law. It is a picture made from people answering questions in a particular place and time. The troubling part of this new picture is not that age suddenly protects everyone. It is that youth no longer appears to.