In twenty years, the three societies should experience a sharp contraction in their working-age population and accelerated aging. The old-age dependency ratio, according to United Nations projections, will deteriorate significantly between 2020 and 2050. Faced with this same constraint, the three countries are mobilizing several levers: automation, employment of older workers, labor immigration, and social reforms, with results that already diverge. It is in this gap that the real costs of each trade-off become visible.
The Essentials
- In East Asia, old-age dependency ratios are rising rapidly: according to UN projections, Japan shows a ratio of approximately 0.50 in 2020, rising to 0.70 in 2050 (for ages 65 and over relative to ages 15-64); South Korea would reach 0.81 in 2050; Taiwan approximately 0.24 in 2020 and 0.72 in 2050 (World Population Prospects 2024).
- South Korea records the world’s lowest fertility rate at 0.72 children per woman; a rebound sufficient to quickly reverse aging appears highly unlikely in central projections.
- Japan has developed industrial automation; Taiwan combines labor immigration and multi-tiered reform; South Korea combines all three levers without a clearly stated strategy.
- Current pension systems were designed for more favorable dependency ratios than those projected for 2050.
- The real stakes are political and social: who contributes, at what age, and according to what rights.
Japan Has Moved First, But the Head Start Has a Price
Japan aged first. It now serves as an involuntary laboratory for its neighbors. Its dependency ratio was already approximately 0.50 in 2020, higher than that of its neighbors, and UN projections place it at 0.70 in 2050 (for ages 65 and over relative to ages 15-64). The country responded with two simultaneous levers: automation of its industrial base and extension of working life.
The first lever is real and measurable. South Korea, according to the International Federation of Robotics, has the world’s highest density of industrial robots, with 1,012 robots per 10,000 manufacturing employees compared with 419 in Japan in 2023. In the assembly plants of Toyota or Panasonic, production lines operate with workforces reduced by half compared with the 1990s. Sectoral automation produces productivity gains whose allocation among wages, prices, taxes, and investment depends on corporate and regulatory choices.
The second lever concerns retirement age. The obligation for employers to ensure employment opportunities up to age 65 derives from reforms predating the Kishida government, particularly the reform applicable since 2013; since 2021, employers have been encouraged to ensure work opportunities up to age 70. In 2023, according to the Japanese Statistics Bureau, 9.14 million people aged 65 or older held jobs, a notable figure among OECD countries.
This solution eases pension funds in the short term. In the longer term, in an economy where fifty- and sixty-year-olds occupy positions, younger people have greater difficulty accessing responsibilities and salaries conducive to starting a family.
South Korea, an Extreme Case Without a Visible Strategy
South Korea concentrates the sharpest contradictions. Its fertility rate of 0.72 children per woman remains the lowest among OECD countries. Fertility has declined sharply over ten years, with a brief increase between 2013 and 2015. At this rate, South Korea’s population could be cut in half by the end of the century, according to UN demographic models.
South Korea’s dependency ratio will reach 0.81 in 2050, the highest among the three countries studied. With this ratio, approximately 1.2 working-age people would remain per person aged 65 or older. South Korean pension systems were built in the 1980s and 1990s, when the ratio was below 0.2, an era when the country had five workers for each inactive person. This gap between the promise made and the available productive base is the heart of the problem.
Seoul has attempted several pro-natalist policies over the past twenty years; despite these measures, fertility reached a new low in 2023. Labor immigration remains regulated and politically sensitive, but it is not blocked; South Korea raised its admission quotas for E-9 workers to 165,000 for 2024. Robotics is advancing, notably in chaebols like Samsung or Hyundai, but without the public-private coordination that characterizes the Japanese model. South Korea pursues multiple policies within an officially stated demographic strategy, even if their coordination and effectiveness can be debated.
Taiwan Attempts the Combination
Taiwan and South Korea both have structured systems for admitting foreign workers, concentrated in sectors exposed to aging: home care, construction, agriculture, small-scale industry. Taiwan already employs several hundred thousand foreign workers.
This targeted immigration policy carries political and social effects. It exerts pressure on wages of low-skilled local workers. Integration of migrant workers into the pension system remains incomplete: depending on status, country of origin, and applicable agreements, some migrants may be excluded from the standard scheme or see their rights limited to repatriation. These tensions are documented and subject to debate in all three countries.
But Taiwan combines immigration with a multi-tiered pension system reform; the Taiwanese system layers public, insurance-based, and occupational schemes with distinct eligibility rules. This layered architecture offers some flexibility to adjust parameters. In 2024, according to OECD pension data, Taiwan showed an effective replacement rate (the share of final salary received as pension) slightly lower than Japan’s, but its system was judged more adjustable in the short term. Taiwan’s economy shows stronger growth dynamics than those of Japan and South Korea.
The Real Cost of Each Trade-off
Comparing the three trajectories makes it possible to name what each choice concretely implies. Automation produces productivity gains whose translation into pension contributions is not automatic. A robot does not contribute to retirement. Without an explicit collection or allocation mechanism, automation gains do not constitute a dedicated resource for financing pensions, though they may contribute indirectly through taxes or other levies. The central political question—taxing productivity gains from machines to feed pension funds—remains unresolved in both Japan and South Korea.
Extension of working life, meanwhile, generates additional contributions, but it requires companies to maintain positions suited to older workers, and requires these workers to remain in good health. This is far from assured in economies where presenteeism and overwork remain cultural norms.
Immigration can produce additional resources for pension systems if immigrant workers are integrated into contribution systems. Exclusion from contribution schemes limits revenues and the protections associated with these schemes, without necessarily eliminating all tax benefits of migrant employment. Taiwan’s model remains fragile on this point. On pension parameter reform—contribution rates, age, benefit levels—all three countries face the same political obstacle: today’s workers do not want to pay more for uncertain future pensions, and current retirees do not want to see their benefits reduced. This blockage is documented in several OECD countries, and the modes of political decision-making on pensions weigh as much as the technical parameters themselves.
The Window from 2030 to 2050: What Signals Already Indicate
UN projections for 2050 are projections, not inevitabilities. The underlying trends of the coming twenty years are nevertheless partly determined by cohorts born since 2000. Measurable signals today make it possible to assess whether a country maintains its trajectory or diverges.
Two indicators deserve close attention. The first is the annual change in the active-to-inactive ratio in each of the three countries, compared with reference UN projections. A stabilization of South Korea’s fertility rate at a less extreme level would reduce, though only partially, the deterioration in the dependency ratio by 2050. The second indicator is the effective pension replacement rate. This figure measures what systems actually provide, not what they promise.
If the replacement rate falls without explicit reform, it means the system quietly erodes rights without politically owning the choices.
Two scenarios structure the coming decade. In the first, Taiwan and Japan combine automation with better-coordinated migrant flows, with Japan gradually increasing admission of foreign workers. Their active-to-inactive ratios improve relative to the most pessimistic trajectories, below current levels but with adjustments to pension promises. In the second, South Korea maintains a fragmented approach: very low fertility, managed immigration, ongoing automation, creating tensions in financing current social commitments.
Between these two trajectories, levers exist. Demographic reserve funds, fed now from productivity gains tied to automation and explicitly directed toward pension systems, would allow for smoothing the shock. The model exists: Norway created a sovereign wealth fund fed by oil revenues; the logic is transposable to robotic gains. Taiwan approaches this in its multi-tiered logic. Japan is beginning to debate it.
South Korea is still at the diagnostic stage. The accelerated aging of East Asian societies also affects their care systems, with consequences for families that merit being read in parallel: in France, the shortage of caregivers already transfers risk to families, a precedent that illuminates what the Asian model seeks to avoid.
One avenue remains underexploited in all three countries: accelerated integration of immigrant workers into formal labor markets, with pathways for qualification recognition and full social rights. A Filipino caregiver in Tokyo who contributes for fifteen years but cannot claim benefits for lack of permanent residence is a lost asset for the system. To contribute to pension revenues, immigrant workers must be affiliated and contributing; portability and scope of rights are additional social protection choices. This is the blind spot common to all three models.
Demography as Constraint, Not Destiny
What Tokyo, Seoul, and Taipei have in common is less demographic than political. UN figures describe a physical constraint: fewer births, greater longevity, fewer workers per retiree. But pension commitments, contribution thresholds, retirement ages, and immigration rules are choices. They can be reformed. They have been—France, Sweden, the Netherlands have all adjusted their pension systems in response to comparable demographic shocks, with different results depending on the decision-making modes adopted.
East Asia adds a specific difficulty: the three countries are aging very rapidly in societies where intergenerational cohesion still relies largely on family rather than state. This dependence on family solidarity is a resource in periods of growth; it becomes a burden when workers are fewer and retirees live longer. Robotics can compensate for part of the labor shortage. Immigration can bring additional contributors. Parameter reform can redistribute effort between generations.
Each of these levers presents its own limitations; their combination requires political decisions that the three governments have yet to clarify. The coming years will measure the capacity of East Asian economies to arbitrate between workers and retirees, between nationals and immigrant workers, between automation gains and aging finance.
Sources
- United Nations, World Population Prospects 2024
- OECD, Data on Pensions and Retirement Systems in Asia (OECD Pensions at a Glance)
- International Federation of Robotics (IFR), data on robot density by country
- Japanese Ministry of Labor (厚生労働省), statistics on senior employment
- Ministry of Finance of the Republic of Korea, data on pro-natalist policy spending
- Executive Yuan of Taiwan, annual report 2026 on migration flows
