The Early Retirement Paradox

In 2009, the Chinese government introduced the New Rural Pension Scheme, a formal pension programme designed to alleviate poverty among elderly rural residents. The intentions were entirely benevolent. Provide financial security for older workers, allow them to stop working and improve their quality of life in later years.

Researchers Plamen Nikolov and Alan Adelman at Binghamton University analysed the programme’s effects using cognitive surveys from the Chinese Health and Retirement Longitudinal Survey. They found that participants in the pension scheme showed significant negative effects on cognitive functioning. The most striking finding was the impact on delayed recall, a measure widely implicated in neurobiological research as an important predictor of dementia. Participants also reported lower rates of volunteering and social interaction compared to non-beneficiaries. The researchers found that the increase in social isolation was strongly associated with faster cognitive decline. (https://www.sciencedaily.com/releases/2019/10/191029131506.htm)

“We were surprised to find that pension benefits and retirement actually resulted in reduced cognitive performance,” Nikolov said.

The programme had done exactly what it set out to do. It had provided financial security and freed people from the obligation to work. But in doing so, it had removed the primary structures through which those people maintained social connection, cognitive engagement and the sense of contributing something to their communities.

This is the early retirement paradox.

Why The Paradox Matters For AI

The early retirement paradox is a finding about what human beings need from structured, purposeful activity and what happens when that activity is removed, even when the material conditions of life are secure.

This matters enormously for how we think about AI-driven displacement.

The dominant response to job displacement, Universal Basic Income (UBI), retraining programmes, income support assume that if the material conditions of life are maintained, people will be fine. The early retirement evidence suggests otherwise.

The people displaced by AI automation will not, in the main, be people who have chosen to stop working. They will be people who are removed from work often mid-career and without warning from roles around which they have built significant professional identity. It is important to note that the evidence on involuntary retirement and forced job loss is consistently worse than the evidence on voluntary retirement. The psychological and cognitive consequences of having work taken from you are substantially more severe than the consequences of choosing to leave it.

A 2025 study published in Scientific Reports by researchers examining the relationship between retirement and emotional memory found that retired older adults with greater depressive symptoms showed enhanced negative and impaired positive memory, a pattern not found in working older adults with equivalent depression levels. The researchers noted that engaging in mental stimulation from one’s occupation can help protect cognition in late life and that when this form of mental stimulation ends, cognitive performance may begin to deteriorate, in line with the “use it or lose it” hypothesis. (https://www.nature.com/articles/s41598-025-87333-9)

This is a direct warning about what AI-driven displacement, at scale and pace, will do to the cognitive health of the people it affects. However the research on retirement and mental health is nuanced which means it matters how we interpret the early retirement paradox in the context of AI.

The Voluntary Versus Involuntary Distinction

A 2023 systematic review published in Global Health Research and Policy examined the association between retirement motivation and mental health outcomes. The findings were consistent with a broader body of research: retirement based on aspirational motivations, retiring because you want to, toward something, significantly reduced depression. Retirement under negative circumstances, redundancy, health problems, organisational restructuring, or being forced out, was associated with deterioration in mental health. (https://ghrp.biomedcentral.com/articles/10.1186/s41256-023-00320-3)

Think of it this way, the people that retire early in order to fill their days with travel, learning skills, enjoying their families or have some framework for how to design their lives is associated with good outcomes. Whereas leaving the workforce without an alternative is catastrophic for your mental health.

The distinction is crucial as the people who will lose their jobs to AI automation are almost entirely in the negative circumstances category. They are not choosing to stop working, rather they are being removed.

In 2022 a review was published in De Economist which surveyed the broad literature on retirement, mental health and cognitive ability. It found that cognitive functioning exhibits a clear lifecycle pattern of general deterioration over older ages, and that retirement accelerates this deterioration beyond what age alone would produce, particularly for men and for those in lower socioeconomic groups. (https://link.springer.com/article/10.1007/s10645-022-09410-y)

What the research consistently finds is that the key moderating factor, the variable that determines whether the transition out of work produces flourishing or deterioration, is whether the retirement preserves or destroys the non-financial benefits that work had provided; social connection, cognitive stimulation, structured activity or a sense of contributing something that matters.

When those things are preserved through other means, the transition can be navigated successfully. When they are not, the damage is real, measurable and, in some cases, irreversible.

The early retirement paradox exposes a significant gap in the policy response to AI-driven displacement. The current policy response to AI displacement is being designed as an economic response to what is, in significant part, a psychological and cognitive challenge and unfortunately the early retirement paradox is the clearest evidence that getting this wrong will produce measurable, lasting harm.

What A Responsible Response Would Look Like

Shifting gears to what we actually need, the first thing would be to take the cognitive dimension of the loss seriously by investing in activities, programmes and community structures that provide the mental stimulation and challenge that work provided. This should be characterised by the deliberate cultivation of the cognitive engagement that prevents deterioration rather than simply retaining for roles.

From the social perspective, the evidence from the Chinese pension study showed that increased social isolation was a significant driver of cognitive decline. Therefore, greater investment in community infrastructure that can provide the regular, purposeful human contact that work provided is essential.

Additionally a distinction is required between voluntary and involuntary transitions, recognising that the people displaced by AI are not in the same psychological position as people who choose to retire, and designing support accordingly. The evidence that involuntary transitions produce substantially worse outcomes than voluntary ones should shape the urgency and intensity of the response.

Finally, a reasonable response would need to abandon the assumption that income replacement is sufficient. The early retirement paradox is the clearest possible demonstration that it is not. While financial security is necessary, on its own it is insufficient.

What You Can Do Now

The policy response may be inadequate and slow but that does not exclude individuals from developing strategies to protect themselves from what seems to be an inevitable change.

The voluntary versus involuntary research shows that cognitive outcomes can be shaped by whether you have something to move toward or not.

The people who navigate the transition well will be those who have already begun building the structures, the social connections, the cognitive engagements, the sources of purpose and identity, that work had previously provided.

This means the most valuable thing you can do right now, before the disruption arrives, is to start building those structures deliberately rather than waiting for circumstances to force you to.

Build social connections that exist outside of your professional identity. Relationships that would survive the loss of your job title. Communities that value you for who you are rather than what you do.

Invest in cognitive engagement that is not dependent on your current role. Learning for its own sake. Skills that challenge you in domains beyond your professional expertise. Creative or intellectual pursuits that will still be there when the work changes.

Develop sources of purpose and contribution that do not live entirely inside your career. Voluntary work, civic participation, mentoring, creative practice, activities through which you experience the sense of mattering that the research consistently identifies as essential to human flourishing.

The early retirement paradox is a warning, but also an invitation. The people who navigate what is coming best will be the ones who started building their own structures early enough that the transition so that when it comes they are already moving towards something rather than away from everything.

So answer this, honestly while you still have the luxury of time to do it thoughtfully:

Who are you, if not what you do?

The AI Transition As An Involuntary Retirement At Scale

The most important implication of the early retirement paradox for the AI transition is that the AI transition is, for a significant proportion of the people it will affect, an involuntary retirement at scale. A forced removal from the structured, purposeful, socially connected activity around which adult life has been organised.

The evidence from retirement research tells us that when it happens to individuals, it results in; cognitive decline, social isolation, depression and deterioration in the non-financial goods of life that work had provided.

Deindustrialisation tells us when it happens to communities, it results in; deaths of despair, generational damage and the hollowing out of the social and civic fabric that purposeful collective work had sustained.

The AI transition threatens to produce both of these outcomes simultaneously at a scale and pace, making every previous episode of mass job displacement look, if not small, then at least more contained.

The early retirement paradox is a warning about what happens when the structures that sustain human cognitive and social life are removed without adequate preparation for what replaces them.

That preparation has not begun.

Next week: “The Lottery Winner Studies”. If money is not the answer, what is? The research on people who win enough to never work again tells us something important about the relationship between financial freedom and human flourishing.