In a world where retirement is often seen as a gateway to freedom, new research reveals that the impact of early retirement on cognitive health isn’t as straightforward as one might expect. This study, focusing solely on men, explores how early withdrawal from the workforce may relate to cognitive decline—a topic that carries significant implications as our population ages.
Yves here. It’s disheartening yet somewhat expected that this study on the effects of early retirement on cognitive decline exclusively examined men. It seems the analysis has made commendable efforts to eliminate men who left their jobs early due to mental health issues or job performance problems—factors that might indicate early signs of dementia.
As someone residing in a predominantly male retirement community—where many are drawn by favorable weather, cost-effective golfing, and sailing, but largely for companionship with Thai women—I find it intriguing. My YouTube feeds brim with advice for this demographic, revealing a common misconception among many: that retiring in an exotic locale would bring about a life-changing transformation. Over time, however, that illusion appears to wear thin.
The study hints at, but does not explicitly address, how men often derive their identities from their work, leaving them feeling lost when they lose the community and status that comes with employment. While it’s often assumed that women are naturally more social and thus less vulnerable to these feelings—an assertion I contest as an introvert—it raises an important question. How much of the cognitive advantages associated with sustained employment stem merely from having a routine and feeling needed, as opposed to engaging in mentally or physically challenging tasks?
To put it simply, should societies take more proactive measures to ensure those who are willing and able to work remain engaged, and also encourage meaningful volunteering? Activities like serving at food banks, reading to the visually impaired, assisting individuals with disabilities, and mentoring children can be immensely fulfilling, fulfilling interpersonal needs while potentially enhancing mental wellness. Furthermore, engaging in altruistic endeavors has been shown to significantly lift one’s happiness.
By Noah Kouchekinia, PhD candidate University of California, Irvine; David Neumark, Distinguished Professor of Economics, University of California, Irvine; and Tim Bruckner, Professor and Chancellor’s Fellow, Wen School of Public Health, University of California, Irvine. Originally published at VoxEU
Dementia currently affects about 6 million Americans, serving as the leading cause of neurological disability and ranking as the third leading cause of overall disability among older adults (those over 65). Research indicates that employment near retirement age can mitigate the risk of cognitive decline, a pivotal precursor to dementia (Coe et al. 2012, Bonsang and Perelman 2012, Celidoni et al. 2017, Mazzonna and Peracchi 2017). The suggested link between work and the delay of cognitive decline implies substantial social advantages of higher employment rates among older individuals.
However, there remains a gap in our understanding of whether mid-life employment—when cognitive decline begins—can postpone cognitive deterioration. While severe cognitive impairment (and the onset of dementia) usually occurs after the age of 65, significant cognitive decline can be detected as early as 40 (Salthouse 2009, Singh-Manoux et al. 2012, Yang et al. 2024). Over 32% of adult men aged 51-64 in the U.S. are currently not participating in the workforce. Our research (Kouchekinia et al. 2026) aims to determine the causal relationship between employment prior to retirement and cognitive decline.
With aging populations, policymakers are increasingly focused on promoting employment beyond traditional retirement ages. In the U.S., motivational factors include the sustainability of Social Security. Given the considerable societal costs related to dementia and cognitive decline, evidence linking employment to cognitive health at pre-retirement ages could shift the cost-benefit analysis concerning policies aimed at boosting employment in this demographic—such as revisions to Social Security Disability Insurance (Maestas 2019) or broader hiring incentives—while further research on cognitive employment benefits for specific target populations is warranted.
The advantages of increasing employment participation among pre-retirement individuals may extend beyond cognitive improvements. Potential benefits include longer work engagement, reduced dependence on disability insurance, and enhanced retirement security through delayed Social Security claims and a larger savings accumulation from extended employment. Moreover, since evidence indicates that an uptick in employment during pre-retirement years helps delay cognitive decline, this suggests additional benefits from policies that encourage later retirement—often justified for different fiscal reasons like strengthening public pension systems. Critically, the ability of such policies to increase employment at younger pre-retirement ages is likely more substantial than at those very close to retirement age, as employment rates are typically higher, and public pensions are not yet accessible to support non-employment.
Prior Work
Descriptive studies reveal a correlation between longer working life and reduced cognitive decline (e.g., Dufouil et al. 2014, Leist et al. 2013, Lupton et al. 2009, Roberts et al. 2011). However, these studies have not established causation because unmeasured mental health factors—like negative mental health events—may influence decisions regarding when to retire or may lead to below-par work performance prior to job loss. Researchers have therefore sought to exploit quasi-random changes in employment to clarify the causal effects of continued work.
Most empirical studies with more rigorous causal designs largely concentrate on policy shifts concerning retirement benefits or Social Security eligibility, or other external alterations impacting employment around retirement ages—such as discontinuities caused by retirement and pension policies (e.g., Atalay et al. 2019, Bonsang et al. 2012, Celidoni et al. 2017, Coe et al. 2012). The consensus indicates that prolonged employment near retirement age can delay cognitive decline, supporting the hypothesis that stimulating cognitive and social activities in the workplace can indeed slow this decline (Salthouse 1991).
Because these studies are based on potentially exogenous retirement policy changes, they have a narrow focus, primarily examining individuals in their mid to late 60s. It remains to be seen whether employment at earlier pre-retirement ages can similarly delay cognitive decline.
Our Approach
Our research leverages a different source of exogenous employment variation compared to studies employing retirement reforms or sharp policy changes. We utilize 12 waves of the Health and Retirement Study (HRS) dataset from 1996 to 2018. The HRS consists of a substantial number of older adults, comprehensive data on employment histories, and detailed trajectories of cognitive decline across diverse geographic regions.1
Our primary research method utilizes County Business Patterns data to construct an indicator of employment shifts at the community level (commuting zones, or CZs).2 We predict local labor demand changes based on initial variations in the industry composition across CZs, weighting national industry-specific growth rates to anticipate shifts in local labor demand. For instance, if a CZ has a significant proportion of its workforce in the automobile industry, and that industry experiences a nationwide decline, we can predict a negative shock to employment demand in that CZ. We utilize these predicted changes to generate variations in older workers’ employment that are independent of other factors potentially influencing both their employment status and cognitive function. This method relies on ‘exogenous’ changes in employment instead of, for instance, cognitive decline prompting exits from the workforce—the very sort of variation that could lead to misleading conclusions regarding the positive effects of employment on cognitive health.
Considerable geographic variation in employment at the CZ level supports the validity of utilizing local demand shocks as an instrumental variable. Figure 1, Panel A, illustrates this for men between ages 51-64. We concentrate on this demographic since the predicted local labor demand changes effectively illustrate actual employment shifts for them. The varying shades depict geographic differences in employment changes and highlight significant employment drops for 51-64-year-old men in various southeastern areas. Panel B showcases the predicted shifts in the employment-to-population ratio across CZs, demonstrating a correlation with realized employment changes—this is the foundation of our empirical strategy.
Figure 1 Change in (employment/population) by commuting zone
A) Men ages 51-64
B) Shift-share instrument prediction
The Evidence
Our research indicates that declines in employment driven by local labor market shocks are linked to cognitive decline among men aged 51-64. We can visually represent this phenomenon effectively. Among HRS participants in our study, cognitive performance exhibits a gradual decline with age. Figure 2 illustrates the trajectories of the global (Langa-Weir) cognitive score as segmented by CZs categorized into employment growth quartiles (from lowest to highest). Respondents from CZs in the lowest quartile of employment growth begin their early 50s with comparable cognitive scores to other quartiles. However, as time progresses, a significant disparity emerges, particularly as respondents approach potential retirement ages. By their 70s, individuals in the bottom quartile display a cognitive score full point lower compared to those in the top quartile of employment growth, suggesting that work influences cognitive decline related to local employment variations.
Figure 2 Cognitive trajectories (global cognitive score) based on quartiles of CZ employment growth
Our estimates identifying the causal relationship between employment and cognitive health affirm this finding: higher employment rates during men’s pre-retirement years play a critical role in delaying cognitive decline.3 This effect appears considerable. For context, data from our sample period indicate that the employment rate drop for men aged 51 to 61 (before becoming eligible for Social Security at age 62) is 18 percentage points. If this decline were halved to 9 percentage points, our estimates suggest this adjustment would elevate the average cognitive score by about one-tenth of a standard deviation—nearly equivalent to the average cognitive decline men experience throughout this age span. In essence, reducing the decrease in pre-retirement employment rates by half could effectively counteract the cognitive decline typically experienced during these years.
Conclusion
This study has its limitations, notably the inability to determine which elements of employment or employment loss directly impact cognitive health. We believe the relationship between work and cognition at pre-retirement ages is variably influenced by both the type of work and the individuals involved. Further research would elucidate additional ways in which both work and non-work activities might help in delaying cognitive decline. Moreover, it is prudent to remain cautious about deriving strong policy recommendations from limited evidence; as highlighted in our findings, research on the causal impacts of work on cognition at pre-retirement ages is still in its initial stages. Nonetheless, our results—particularly if confirmed by additional studies—have clear implications for policy. Federal initiatives aimed at boosting work engagement for pre-retirement individuals could not only reduce reliance on disability insurance and improve retirement security but also foster healthier aging by delaying cognitive decline.
_________
- We obtained restricted-use data on county of residence and industry of employment (RDA #2023-095). The conditions for utilizing this data limit the geographic granularity at which we can report statistics.
- This uses a Bartik-style shift-share instrumental variable (Bartik 1991).
- The paper includes multiple auxiliary analyses that confirm the robustness and reliability of these estimates for men aged 51-64. However, this research framework is less effective at determining whether similar results apply to women in the same age range, or older individuals, since local labor demand shocks do not predict their employment changes as effectively—likely because many older individuals retire for reasons unrelated to local demand, and jobs held by women often exist in sectors with less local labor demand variability.
See original post for references
