Working to 70 — What the Oldest Workforces Have Learned

The short answer to the question this report examines is that working longer is already happening, it is economically valuable, and it is far less evenly available than governments assume. Across the OECD, employment among people aged 60 to 64 has risen by 21.8 percentage points to 56.5% in 2024 [2], and Japan now employs 53.6% of everyone aged 65 to 69 [3] ✓ Established. Yet the capacity to work into one's late sixties is distributed by health, education, occupation and the prejudice of hiring managers — not by birthday. Policies that raise the pension age uniformly therefore transfer the cost of demographic change onto the people least able to carry it. The rest of this report sets out the evidence for each part of that claim: the demographic arithmetic, the economics, the discrimination data, the health and skills constraints, and what the countries with the oldest workforces — Japan, Korea, Singapore, Denmark, Germany and others — have learned by going first.

Begin with the arithmetic, because it is unforgiving. The OECD's Pensions at a Glance 2025 projects that the working-age population of its member countries will fall by 13% over the next 40 years, and by more than 30% in Japan, Korea, Italy, Spain, Greece, Poland, Slovakia and the Baltic states [1] ✓ Established. In Korea the old-age dependency ratio is set to rise by almost 50 points [1]. Statistics Korea projects that the national population will shrink from 52 million to 36.2 million by 2072, that 47.7% of residents will then be aged 65 or over, and that the total dependency ratio will climb from 42.5 to 118.5 [4] ✓ Established. In other words, by the early 2070s Korea would have more dependants than people of working age. These are not distant hypotheticals: Korea formally became a super-aged society on 23 December 2024, when 10.24 million residents, 20% of the population, were 65 or older [5].

The global picture follows the same curve, only later. The United Nations' World Population Prospects 2024 projects that world population will peak at around 10.3 billion in the mid-2080s, and that by the late 2070s people aged 65 and over — 2.2 billion of them — will outnumber children under 18 [6] ✓ Established. Population has already peaked in 63 countries, including China, Germany, Japan and Russia, and that group is projected to shrink by 14% over the next 30 years [7]. The demographic transition that began in Europe and Japan is now a feature of middle-income economies too, which is why China in September 2024 legislated its first increase in the statutory retirement age since the 1950s [8]. A shrinking society has three levers to keep output and public finances stable: more workers from abroad, more output per worker, or more years of work per person. Most governments now pull all three, but the third is the one written into law.

That third lever is visible in the retirement ages already legislated. The OECD's normal retirement age — the age at which a full pension can be drawn after a full career — stood at 64.7 years for men and 63.9 for women retiring in 2024; for people who entered the labour market in 2024 it is projected to reach 66.4 and 65.9 [9] ✓ Established. Some countries go much further. Because Denmark, Estonia and Italy link pension ages to life expectancy, the OECD projects future normal ages of 74 in Denmark, 71 in Estonia and 70 in Italy [10]. Even so, the OECD calculates that legislated increases offset only slightly more than 40% of the projected gain in life expectancy at 65, which for men rises from 18.5 to 22.7 years [10] ◈ Strong Evidence. Retirement, in other words, will keep getting longer even as it starts later.

The OECD Secretary-General, Mathias Cormann, framed the problem at the launch of the 2025 report in deliberately structural terms: population ageing is a key structural challenge across OECD countries, with significant economic, fiscal and social implications [1]. The fiscal side is easiest to quantify. In the United Kingdom, the Office for Budget Responsibility estimates that the move from a State Pension age of 66 to 67, phased in between 2026 and 2028, will reduce borrowing by 10.5 billion pounds in 2029-30, almost all of it because about 820,000 fewer 66-year-olds will be drawing a pension [11] ✓ Established. Every one-year rise in a pension age is, in budgetary terms, a large and immediate saving. That is why it is the instrument of choice.

The economic side is harder, because a pension age is only a signal; it does not by itself create a job, a healthy body or an employer willing to hire. The International Monetary Fund made the point plainly in its April 2025 World Economic Outlook, warning that raising the statutory retirement age may not yield the desired impact on the effective retirement age if older workers are not physically or cognitively able to remain active [12] ✓ Established. The gap between the legal age and the age at which people actually stop working is where the real policy question lives. The remaining sections examine that gap: who is already working longer, what it is worth, who is being kept out, and what the oldest workforces have learned about closing it.

The headline trend is stronger than public debate suggests. According to the OECD, the employment rate of 55-to-59-year-olds rose by 17.1 percentage points to 75.7% in 2024, while that of 60-to-64-year-olds rose by 21.8 points to 56.5% [2] ✓ Established. The average OECD employment rate for 65-to-69-year-olds reached 26.4% [13]. In the European Union, the employment rate of people aged 55 to 64 hit a record 65.2% in 2024 [14] ◈ Strong Evidence. Sweden, long the benchmark, reached about 78.1% in the same age band, and Germany 75.15% [15] [16] ✓ Established. Countries with a retirement age of 67 for both sexes — Denmark, Iceland, the Netherlands and Norway — are among those with the highest employment rates at 60 to 64 [2]. The pension age, in other words, is not merely a fiscal parameter: it anchors when people and employers expect working life to end.

Japan sits at the far end of the distribution. In 2024 the number of employed people aged 65 and over reached a record 9.3 million, the 21st consecutive annual increase [3] ✓ Established. The employment rate was 53.6% at ages 65 to 69, 35.1% at 70 to 74 and 12.0% at 75 and over, all record highs; across everyone aged 65 or over it was 25.7% [3]. A decade earlier, in 2013, only 38.7% of 65-to-69-year-olds were in work [17]. Older people now make up 29.4% of Japan's population [3], and the country has in effect reorganised its labour market around the expectation that most people will keep working into their late sixties. That reorganisation, as Section 6 shows, has come at a price in wages and job quality.

The United States presents a different shape. Labour force participation among Americans aged 65 and over stood at 19.1% in 2025 — nearly one in five — up from 12.9% in 2000 but below the 20.2% recorded in 2019 [18] ✓ Established. The pandemic produced what the Federal Reserve Bank of St. Louis described as excess retirements: about 2.4 million more Americans retired than pre-pandemic trends predicted, with the gap peaking near 2.8 million in late 2022 [19]. By late 2024 and early 2025, however, the share of excess retirements had fallen below zero, and the St. Louis Fed concluded that labour market conditions and government transfers explained the bulk of the increase [20] ◈ Strong Evidence. The fastest-growing segment is the oldest: workers aged 75 and over made up 1.2% of the US labour force in 2023 and are projected to reach 2.1% by 2033 [21].

The United Kingdom illustrates the opposite risk: that ill health pulls older people out of work faster than pension policy pushes them in. The Office for National Statistics found that more than 2.5 million people were economically inactive because of long-term sickness in 2023, an increase of more than 400,000 since the pandemic [22] ✓ Established. Among inactive 50-to-64-year-olds, 46% reported five or more health conditions, up from 41% in 2019 [22]. Britain is raising its State Pension age from 66 to 67 between May 2026 and March 2028, and a third statutory review, launched in July 2025, is examining the move to 68 [23]. The British case shows why employment rates and pension ages cannot be read in isolation: a higher pension age without the health to match produces not more workers but more people waiting, on lower incomes, for a pension that has moved further away.

Three patterns stand out from the cross-country data. First, older-worker employment responds strongly to the pension age itself: when the age at which benefits begin rises, employment just below that age rises with it, as the evidence from Britain in Section 3 shows. Second, the countries with the highest employment at 60 to 64 combine a late pension age with strong labour markets, flexible work arrangements and active adult-learning systems — the Nordic model rather than the Japanese one. Third, the gains are concentrated among the healthy and the educated. The rest of this report tests each of these patterns against the evidence, starting with what working longer is actually worth.

• Japan: 53.6% of people aged 65 to 69 in work in 2024, a record [3]
• Sweden: about 78.1% of people aged 55 to 64 in work, among the highest in the OECD [15]
• Germany: 75.15% of people aged 55 to 64 in work in 2024 [16]
• European Union: a record 65.2% of people aged 55 to 64 in work in 2024 [14]
• United States: 19.1% of people aged 65 and over in the labour force in 2025 [18]

The most important recent contribution to this debate is the IMF's April 2025 World Economic Outlook chapter on the silver economy. Drawing on data from 41 advanced and emerging economies, it found that, on average, a person who was 70 in 2022 had the same cognitive ability as a 53-year-old in 2000 [12] ✓ Established. On physical frailty the improvement was similar, with a 70-year-old in 2022 matching a 56-year-old in 2000 [12]. The chapter then linked those health gains to behaviour: a decade's worth of cognitive improvement is associated with a rise of about 20 percentage points in the likelihood of remaining in the labour market, about six more hours of work a week and 30% higher earnings for those employed [12] ◈ Strong Evidence. The phrase the IMF chose to summarise the result — the 70s are the new 50s — has since been widely quoted, often without the caveats attached to it.

The macroeconomic stakes are large. The IMF estimates that demographic change explains almost three-quarters of the projected slowdown in global growth, which it puts about 1.1 percentage points below its 2016-18 average; without the gains from healthier ageing, the shortfall would be 1.5 points [12] ◈ Strong Evidence. Earlier private-sector estimates point the same way. PwC's Golden Age Index calculated in 2018 that raising employment among workers aged 55 and over across the OECD to Swedish levels would add about US$3.5 trillion to long-run OECD GDP [24] ◈ Strong Evidence. Such figures depend on assumptions about who could work and at what productivity, and should be treated as orders of magnitude rather than forecasts. But they agree on direction: in an ageing economy, the employment rate of people over 55 is one of the largest single determinants of output per head.

The public-finance evidence is stronger still, because it comes from real reforms. When the United Kingdom raised the State Pension age for women from 60 to 61, the Institute for Fiscal Studies found that the employment rate of 60-year-old women rose by 7.3 percentage points — about 27,000 more women in work — and that their husbands' employment rose by 4.2 points, strengthening public finances by about 2.1 billion pounds [25] ✓ Established. The IFS economist Jonathan Cribb noted that the size of the effect suggested that things other than pure financial incentives were at work: the pension age acts as a social norm, a signal of when stopping is expected [25]. That signalling effect is precisely what makes pension-age reform powerful — and what makes it blunt.

The most persistent obstacle to working longer is the belief that older workers are less productive. The best micro-level evidence does not support it. Axel Börsch-Supan and Matthias Weiss studied errors made by work teams on the assembly line of a large car manufacturer, correcting for the fact that less productive workers tend to leave, and found that productivity in this plant — typical of large-scale manufacturing — does not decline at least up to age 60 [26] ◈ Strong Evidence. Older workers made slightly more errors, but far fewer severe ones, because experience taught them which mistakes mattered. On a production line, accumulated judgement offset any slowing of physical speed. The result has become one of the most cited findings in the economics of ageing precisely because it contradicts what managers expect.

At the level of whole economies, the picture is more mixed ⚖ Contested. Nicole Maestas, Kathleen Mullen and David Powell estimated that a 10% increase in the share of the population aged 60 and over reduces growth in GDP per capita by 5.5%, with roughly two-thirds of the effect coming through slower growth in labour productivity rather than fewer workers [27]. That finding does not show that individual older workers are less productive; it may reflect slower adoption of new technology, lower investment or changing patterns of demand in ageing regions. But it is a warning against assuming that a longer working life is automatically a more productive one. The reconciliation, consistent with both studies, is that older workers remain productive in jobs that use their experience — and that economies which fail to invest in their skills and health forgo much of the gain.

The strongest evidence on age discrimination comes from field experiments, in which researchers send employers matched fictitious applications that differ only in age. The largest of these, by David Neumark, Ian Burn and Patrick Button, submitted more than 40,000 applications for jobs in 12 American cities and four occupations [28] ✓ Established. After correcting for biases that afflicted earlier studies, they found robust discrimination against older women, particularly those close to retirement age, and weaker but present evidence against older men [28]. Neumark summarised the pattern bluntly in 2019: older applicants never do better, and often do worse, and women face age discrimination starting in their forties [29]. A Swedish experiment by Magnus Carlsson and Stefan Eriksson, using more than 6,000 applications for applicants aged 35 to 70, found callback rates falling substantially from about age 40 and more steeply for women than for men [30] ✓ Established.

What employers say they believe helps explain the results. In the Swedish experiment, employers who were later surveyed cited stereotypes about older applicants' ability to learn new tasks, their adaptability and their ambition [30]. Workers recognise the same pattern from the other side. AARP's 2026 survey found that 64% of American workers aged 50 and over have seen or experienced age discrimination at work, rising to 74% among Black workers and 67% among women [31] ✓ Established. Some 90% believe age discrimination is common, and 33% report having been assumed to be less comfortable with technology [32]. Among people aged 50 and over who planned to change jobs in 2025, 74% believed their age could stand in the way of being hired [33].

Formal complaints are only the visible fraction of the problem. The US Equal Employment Opportunity Commission received 16,223 charges of age discrimination under the Age Discrimination in Employment Act in fiscal year 2024, a rise of 9.2% on the previous year, according to an analysis by the law firm Seyfarth; the Commission itself filed only seven age-discrimination lawsuits [34] ◈ Strong Evidence. Age discrimination is notoriously difficult to prove because it rarely leaves a written trace: a candidate is described as overqualified, a team is said to need fresh energy, a restructuring happens to remove the oldest staff. Carly Roszkowski of AARP described the result in January 2025: many older workers face age discrimination and are often overlooked for jobs they are qualified for, even though they bring expertise, maturity and perspective [33].

The costs are not only individual. The World Health Organization's Global Report on Ageism, published in March 2021 with three other UN bodies, found that one in two people worldwide hold ageist attitudes towards older people [35] ✓ Established. The report estimated that ageism was responsible for 63 billion US dollars of excess health spending in the United States in a single year, and that employing 5% more Australians aged 55 and over would be worth 48 billion Australian dollars a year to the economy [35] ◈ Strong Evidence. The WHO Director-General, Tedros Adhanom Ghebreyesus, warned that countries could not let age-based stereotypes, prejudice and discrimination limit opportunities to secure the health, well-being and dignity of people everywhere [35].

Discrimination also has an institutional form that is rarely called by that name: the age discount built into pay. In Japan, where firms are legally required to offer continued employment to 65 and to make efforts to do so until 70, re-employment after the company retirement age commonly comes with a steep pay cut. Only 39.6% of Japanese firms keep post-retirement pay above 80% of the pre-retirement level, which means roughly 60% cut it by more than a fifth for the same or similar work [36] ✓ Established. Korea's mandatory retirement age of 60 produces a similar cliff [37]. These systems keep older people in work, but on terms that would be regarded as discriminatory if applied to any other group. They are the price at which employers have accepted the obligation to keep them.

The case for working longer rests on the assumption that longer lives are also healthier lives. On average that is broadly true, but the average hides very large gaps. The World Health Organization's World Health Statistics 2024 reported global life expectancy of 71.4 years in 2021 and healthy life expectancy of 61.9, meaning people live on average about 9.5 years in poor health [39] ✓ Established. Within countries the variation is wider than between them. The Office for National Statistics found that in England, healthy life expectancy for men in the least deprived areas was 70.1 years in 2020-22, compared with 51.1 years in the most deprived — a gap of 19.0 years; for women the gap was 19.7 years [38] ✓ Established. Men in England's poorest areas can therefore expect to lose good health more than 15 years before reaching the State Pension age.

The longevity gap by income is just as stark in the United States. Raj Chetty and colleagues, using tax and mortality records, found that the richest 1% of American men live 14.6 years longer than the poorest 1%, and the richest women 10.1 years longer [40] ✓ Established. The gap is widening: between 2001 and 2014, life expectancy rose by 2.34 years for men in the top 5% of the income distribution but by only 0.32 years for men in the bottom 5% [40]. Across 23 OECD countries, highly educated men aged 65 can expect to live 3.6 years longer than those with low education, and women 2.6 years longer [41] ✓ Established. Because pension-age rises are justified by average longevity gains, and because those gains accrue mostly to the affluent, a uniform increase asks the people who gained least to give up the most.

Health limits are compounded by the physical character of many jobs. A 66-year-old lawyer and a 66-year-old care worker face the same pension age, but not the same body at the end of a working week. The OECD's Employment Outlook 2025 devoted a full chapter to the skills and jobs of older workers in a changing labour market, reflecting a growing recognition that the constraint is often the job rather than the worker [42]. In the UK, 55% of those inactive because of long-term sickness in mid-2022 — about 1.3 million people — were aged 50 to 64 [22] ✓ Established. For this group, a later pension age does not mean more work. It means longer reliance on sickness and disability benefits, or on savings, before the pension arrives.

Skills are the second constraint, and the data on them is newer. The OECD's Survey of Adult Skills 2023, published in December 2024, found average adult scores of 260 in literacy, 263 in numeracy and 251 in adaptive problem solving across participating countries [43]. People aged 55 to 65 scored below those aged 25 to 34 in every domain, by 30 points in literacy on the OECD average, by 55 points in Korea, 26 in England and 17 in the United States [44] ✓ Established. The OECD cautions that the gaps could reflect ageing effects but also differences in the quality and quantity of education and training across generations [44]. Korea's exceptionally wide gap is consistent with that reading: its older cohorts were educated before the country's great expansion of schooling.

The training system then widens the gap it should be closing. Only about 33% of 60-to-65-year-olds took part in training in 2023, compared with more than 50% of 25-to-44-year-olds, and the share learning on the job fell from 62% at ages 25 to 29 to 45% at ages 60 to 65 [42] ◈ Strong Evidence. Earlier OECD data put participation in job-related training at 24% for people aged 55 to 65 against 41% for those aged 45 to 54 [45] ✓ Established. Employers invest less in older staff because they expect a shorter payback; older staff invest less in themselves for the same reason. Yet a worker of 55 facing a pension age of 67 or 70 has 12 to 15 years of working life ahead, longer than the average job tenure of a younger employee. The training gap is a relic of an era when 55 was the beginning of the end.

Artificial intelligence adds a new and ambiguous variable. The OECD finds that exposure to automation is highest among younger workers and falls quickly with age, while exposure to AI more broadly follows an inverted U shape across the life course; it also finds that use of generative AI grew fastest among people aged 55 to 74 in 2024-25, albeit from a low base [46] ◈ Strong Evidence. A survey reported by the Wharton Pension Research Council put generative-AI use among workers aged 50 to 64 at 18%, well below that of workers in their thirties and forties [47]. Some 34% of American job-seekers aged 50 and over said in January 2025 that they worried AI threatened their job security [33]. AI could extend working lives by taking over the physically and cognitively draining parts of jobs — or shorten them by devaluing experience. Which happens depends heavily on whether older workers are trained to use it.

Japan's approach is to place the obligation on employers rather than on pensions. Firms must offer employment to 65, and since a 2021 amendment they must make efforts to secure work opportunities until 70 [48]. The share of firms with measures to secure work to 70 rose from 25.6% in 2021 to 34.8% in 2024 [48] [36] ✓ Established, and small and medium-sized firms have moved faster than large ones, at 32.4% against 25.5% in the 2025 data [48]. The model works in terms of headcount: Japan has the highest employment rates for people in their late sixties of any large economy [3]. Its weakness is job quality. Re-employment typically comes on fixed-term contracts with sharply lower pay, which keeps people working but often in roles below their skills [36].

South Korea shows what happens when the mandatory retirement age and the pension age diverge. Korea's statutory retirement age is 60, while eligibility for the national pension is rising to 65 by 2033, leaving a gap of up to five years without either salary or pension [37] ✓ Established. Around 40% of Koreans aged 65 and over live below the relative poverty line, the highest rate in the OECD and well above the OECD average of about 14.8% [49] ◈ Strong Evidence. The National Human Rights Commission recommended raising the retirement age to 65 in March 2025, and in June 2026 the ruling Democratic Party's committee proposed raising it by one year every two years from 2028, reaching 65 in 2036 [50]. Public support is high, with 88.3% backing a phased rise [51].

The Korean debate also exposes who pays. Ryu Gi-seop of the Federation of Korean Trade Unions described the income crevasse between retirement and pension payments as a structural problem directly tied to the country's elderly poverty rate, the highest in the OECD [37].Employers, represented by Lee Sang-cheol of the Korea Enterprises Federation, argue that a uniform extension would concentrate benefits on large companies and unionised workers with regular contracts, deepening labour market inequality [37] ⚖ Contested. The Korean Confederation of Trade Unions, from the opposite side, described the phased plan as tantamount to abandoning hundreds of thousands of workers to an income gap [50]. Both are right about part of the problem: a higher retirement age protects insiders with secure jobs, while those in precarious work — who need income most — benefit least. Korea's year-end 2026 deadline for a decision will test whether a compromise can address both [37].

Singapore offers the most carefully sequenced model. Under its Retirement and Re-employment Act, the retirement age rose from 63 to 64 and the re-employment age from 68 to 69 on 1 July 2026, on a path to 65 and 70 by 2030 [52] ✓ Established. Employers must offer re-employment to eligible workers up to the re-employment age, giving older staff a legal expectation of continued work without freezing their old contracts. Denmark took the opposite route, legislating the outcome rather than the obligation: in May 2025 its parliament voted 81 to 21 to raise the pension age to 70 by 2040 for people born after 1970, via 68 in 2030 and 69 in 2035 [53] ✓ Established. Denmark has linked its pension age to life expectancy since its 2006 welfare reform [53]. Even so, Prime Minister Mette Frederiksen had questioned the automatic mechanism in 2024, saying people could not simply keep being told to work a year longer [54].

Germany chose incentives over obligation. Its Aktivrente, in force since 1 January 2026, exempts up to 2,000 euros a month of employment income from income tax for people who keep working beyond the statutory retirement age [61] ✓ Established. The scheme excludes the self-employed, civil servants and those in mini-jobs [62]. Finance Minister Lars Klingbeil presented it as a reward: anyone who chooses to work longer should see a real gain in their pocket [61]. Its likely beneficiaries, however, are those already able and willing to work past 67 — typically healthier, better-paid employees — which raises questions about deadweight cost that only evaluation will answer. Sweden, by contrast, has built adjustment into the system itself: its target retirement age, the riktålder, is set at 67 for 2026 to 2031, with earliest withdrawal at 64, and moves with life expectancy [63] ✓ Established.

China and France illustrate the political limits. China's reform, adopted by the National People's Congress Standing Committee in September 2024, raises the retirement age for men from 60 to 63 and for women from 50 to 55 or from 55 to 58 over 15 years from January 2025, while lifting the minimum contribution period from 15 to 20 years by 2039 [8] ✓ Established. It was prompted in part by projections from the Chinese Academy of Social Sciences that the urban pension fund could be exhausted by 2035 [64] ◈ Strong Evidence. France raised its minimum age from 62 to 64 in 2023 despite mass protests [60], then, under Prime Minister Sébastien Lecornu, suspended the reform in late 2025, freezing the legal age at 62 years and 9 months until January 2028 [65] [66] ✓ Established. Lecornu promised that no rise in the age would occur from now until January 2028 [65]. The lesson is not that reform is impossible but that it is reversible when it lacks consent.

• Employer obligation (Japan, Singapore): firms must offer continued work to a set age; high employment, lower pay and job quality [36] [52]
• Automatic link to life expectancy (Denmark, Sweden): predictable but politically contested as ages approach 70 [53] [63]
• Tax incentives (Germany): rewards those already able to work; deadweight risk [61]
• Legislated step increases (China, UK, US): large fiscal savings, uneven employment effects [8] [11]
• Suspension or reversal (France): shows the cost of reform without consent [65]

The oldest objection is that older workers take jobs that would otherwise go to the young. Economists call this the lump-of-labour fallacy, and most of the evidence supports the label. A multi-country NBER project led by Jonathan Gruber, Kevin Milligan and David Wise, covering 12 countries, found no evidence that increasing employment of older people reduces employment opportunities for the young [67] [68] ◈ Strong Evidence. In US state-level data from 1977 to 2011, Alicia Munnell and April Yanyuan Wu at Boston College found that greater employment of older workers was associated with lower youth unemployment and higher youth employment; the theory may sound plausible, they concluded, but the data do not support it [69]. Over time, economies adjust: employed older people spend, save and create demand for other workers.

But the short run can look different ⚖ Contested. Studying Italy's 2011 Fornero reform, which abruptly raised retirement ages, Tito Boeri, Pietro Garibaldi and Espen Moen found that for every five older workers locked into their jobs for an extra year, firms reduced youth hiring by about one full-time equivalent [70]. The effect was concentrated in firms facing financial constraints, where a fixed wage bill meant one more senior salary displaced one junior hire. The two bodies of evidence are reconcilable: in a growing economy with flexible hiring, older workers complement younger ones; in a stagnant one with rigid payrolls, sudden reforms can block the entry of the young. Sequencing and economic conditions matter as much as the reform itself.

The fairness question is less contested on the facts than on the remedy ⚖ Contested. Nobody disputes that life expectancy differs sharply by income and education [40] [41] [38]. The disagreement is over whether this makes pension-age increases unacceptable or merely requires compensating measures — easier access to disability benefits, earlier retirement for long careers or arduous jobs, or more generous pensions at the bottom. Denmark's own debate shows the tension: even a government that legislated retirement at 70 has questioned the automatic mechanism behind it [54]. The IMF notes that socio-economic gaps in health by education, wealth and rural residence have persisted even as average health improved [12] ✓ Established.

The productivity dispute turns on the level of analysis ⚖ Contested. Studies of individual workers in specific settings — the Börsch-Supan assembly line is the best known — find no decline up to 60 [26], while studies of whole regional economies find that ageing populations grow more slowly, mainly through productivity [27]. The cognition dispute is similar. Susann Rohwedder and Robert Willis, comparing countries with different retirement ages, found that retirement had a significant negative impact on the cognitive ability of people in their early sixties, suggesting that retirement can lead to a less stimulating daily environment [71]. But Norma Coe and colleagues, using unexpected early-retirement offers as a natural experiment, found no causal negative effect for white-collar workers and a positive one for blue-collar workers [72]. Whether work protects the mind may depend on what the work is.

Automation and immigration are often presented as alternatives to working longer ⚖ Contested. Daron Acemoglu and Pascual Restrepo found that ageing societies adopt industrial robots faster, but that robots tend to substitute for middle-aged workers aged 36 to 55 rather than for older ones [73]. Automation therefore compensates for a shortage of prime-age workers rather than removing the need for people to work longer. Immigration can offset part of the decline in the working-age population, but in countries such as Japan and Korea, where the working-age population is projected to fall by more than 30% [1], the scale required would be politically and socially unprecedented. None of the three levers — immigration, technology or longer careers — is sufficient alone, which is why the debate over working longer is not about whether, but about how.

The weight of evidence suggests a qualified conclusion. Working longer is feasible for most people in good health and in jobs that use accumulated skill, and it does not, in general, cost young people their jobs ◈ Strong Evidence. But it is not equally feasible for everyone, the labour market does not welcome it, and the costs of uniform reform fall disproportionately on people with shorter and less healthy lives. The contested questions are therefore less about whether people should work longer than about which people, in which jobs, with what protection for those who cannot.

The evidence assembled in this report points to five conclusions. First, working longer is not a future policy but a present reality: OECD employment among people aged 60 to 64 has risen by more than 20 percentage points [2], and Japan employs a majority of people in their late sixties [3] ✓ Established. Second, it is valuable: the IMF estimates that healthier ageing adds about 0.4 percentage points a year to global growth [12], and pension-age reforms produce immediate fiscal savings [11]. Third, the main barrier is often not workers' capacity but employers' willingness, as field experiments on hiring show [28]. Fourth, the capacity to work longer is unequally distributed by health, education and income [38] [40]. Fifth, reforms imposed without consent can be reversed, as France demonstrated in 2025 [65].

The risks of the current policy mix can be assessed against that evidence. A pension age that rises faster than healthy life expectancy for lower-income groups risks pushing people onto disability benefits or into poverty rather than into work. Hiring discrimination means that older jobseekers who lose their positions in their late fifties may never find comparable work again. Training systems remain designed for workers with long careers ahead, not for those of 55 who now have 15 years to go. And the pay discount on re-employment, as in Japan, can make longer careers economically viable for employers but unattractive for workers [36].

The policy response that the evidence supports is broader than a retirement age. It treats the later decades of working life as a stage to be designed for rather than an extension to be imposed. That means enforcing age-discrimination law with the seriousness applied to other protected characteristics, since complaints in the United States reached 16,223 in fiscal year 2024 while the federal enforcement agency brought only seven lawsuits [34] ◈ Strong Evidence. It means mid-career training in the fifties, when 12 to 15 years of working life remain, rather than retraining after redundancy. It means occupational health measures and flexible work that allow people in physically demanding jobs to remain productive, and dedicated routes out of the labour market for those who cannot.

It also means linking pension ages not to average longevity but to evidence on who can actually work. Several countries already allow earlier retirement for long careers or arduous occupations; the evidence on longevity gaps suggests these exceptions are not loopholes but corrections [41] [40]. Germany's tax-free earnings and Singapore's re-employment ladder are early experiments in making longer work attractive rather than merely compulsory [61] [52]. Their results, together with China's long phased transition [8] and Denmark's move to 70 [53], will form the next generation of evidence.

The anti-ageing of the workforce is, in the end, a test of whether societies can match the biological good news of longer and healthier lives with institutions that let people use those years productively and fairly. The IMF's finding that today's 70-year-olds have the cognitive capacity of yesterday's 53-year-olds [12] is genuinely good news. But it describes an average, and pension policy is lived by individuals. For the healthy, educated and well-employed, working to 70 is increasingly realistic and sometimes desirable. For the care worker, the labourer and the woman screened out of hiring at 55, it is a promise the labour market has not yet made. Closing that gap, rather than raising the age, is the real policy task of a shrinking society.