Ageing without a benefit wave

Ageing was supposed to strain the welfare state: more pensioners, fewer workers to carry them. The first half is true. In November 2025, 3.37 million residents lived mainly on a pension, against 2.43 million in 2001. Their share of the population rose from 15.1% to 18.6%.

The second half is not. Every November the CBS counts how many people live on a benefit for every 100 people whose main income is work. In 2025 there were 55.6, of whom 39.6 drew a pension and 16.0 a social security benefit. The peak was 2014, at 63.6. Since 2022 the figure has not moved: 55.1, 54.9, 55.3, 55.6.

A million more pensioners and fewer benefits per worker than ten years ago: that is the puzzle, and the answer is that the denominator grew faster than the numerator.

The number of residents living mainly on work rose from 7.14 million in 2014 to 8.51 million in 2025, a gain of 19.2%. Pensioners grew from 3.08 to 3.37 million over the same years, 9.3%. People on a social security benefit, meaning welfare, unemployment, disability, or the Wajong scheme for young disabled people, fell from 1.46 to 1.36 million. Per 100 workers, the pension figure went from 43.2 to 39.6 and the social security figure from 20.4 to 16.0.

Nearly half the whole population, 47.0%, lived mainly on paid work in 2025. The CBS calls that the highest share in 25 years. The group with neither work nor a benefit, mostly pupils, students, and partners without an income of their own, shrank from 33.5% of the population in 2001 to 26.9% in 2025.

Much of the growth in workers sits above 55. In the labour force survey, 1.15 million people aged 55 to 65 worked in 2010 and 1.88 million in 2025. Among 65 to 75 year olds, 127,000 worked in 2010 and 412,000 in 2025. Together that is more than a million additional older workers in fifteen years.

The state pension age stood at 65 through 2012. It then rose in steps to 66 in 2018, 67 in 2024, and 67 years and three months in 2028. Someone who was sixty in 2010 could draw the state pension at 65. Someone who is sixty in 2025 waits until 67 and three months.

Early retirement disappeared over the same years. In 2006, 18.9% of all 55 to 65 year olds lived mainly on a pension. In 2025 it was 2.9%. Tax support for early retirement schemes had already been wound down, and the generation that still had rights under them has since passed 65.

The result is in the participation figures. Net participation among 60 to 65 year olds went from 36.9% in 2010 to 69.5% in 2025, close to a doubling in a group of 1.21 million people. A growing group works past the pension age as well: 29.5% of 65 to 70 year olds and 10.7% of 70 to 75 year olds.

The growth in the working population therefore did not come only from birth cohorts or migration. It came from an age group that fifteen years ago had largely stopped.

Three things blunt the figure.

First, it is a ratio, not a cost. The 3.37 million pensioners cost more than the 2.43 million of 2001, whatever the ratio does. That a worker in 2025 carries fewer benefit recipients than in 2014 says nothing about what those benefits cost. And the pension figure per 100 workers still sits above its 2001 level of 34.7.

Second, the social security figure has been rising again since 2022, from 15.3 to 16.0 per 100 workers. The stock of unemployment benefits grew 9.5% in 2025 to 191,459 and stood at 199,700 at the end of July 2026. The stability of the total hides a falling pension share and a rising benefit share.

Third, the CBS counts residents whose main income is a benefit, in November, from registers. Someone who works past the pension age and earns more than their pension counts as a worker. Pensioners abroad are not counted. The 2025 figure is provisional. And the labour force survey behind the participation figures changed its design in 2021, so 2010 and 2025 straddle a break in the series.

The rise in participation above 55 is European, and the Netherlands leads it. Eurostat puts the employment rate of 55 to 64 year olds at 52.9% in 2010 and 75.8% in 2025, a rise of 22.9 percentage points. Only Belgium rose faster, by 24.2 points, from a much lower level. The EU-27 went from 44.7% to 66.4%.

Past 65, only Norwegians, Danes, and Swedes work more often than the Dutch. The Netherlands made the largest jump: from 9.5% to 20.8%, 11.3 points, against 11.4 in Denmark and 11.2 in Germany. The expected working life reached 44.0 years in 2025, the longest of the nine and 5.1 years longer than in 2010. Across the EU it is 37.5 years.

The coalition agreement of January 2026 leaned on more labour supply. Its largest item was tying the state pension age one-to-one to life expectancy from 2033, worth €2.8bn a year in the long run. The CPB, the government's forecaster, costed the whole package at 1.3% more structural employment in hours and 1.2% in persons, and attributed that mainly to the pension age link.

On 26 May 2026 the cabinet dropped the pension proposal, in a letter to the unions that also went to parliament. The same letter says the unemployment and disability benefit plans will for now not go ahead in the proposed form. The CPB analysis has not been rerun since, and on 10 September 2026 the pension item still sits in the budget table without cover.

Of the unemployment benefit plans, the entry test is the best costed. A claimant today must have worked 26 of the last 36 weeks; the agreement makes that 42 of 52. UWV, the benefits agency, says in answers to parliamentary questions of 31 August 2026 that about 12% of the people awarded a benefit in 2024 and in 2025 would not have qualified under the new test. The CBS earlier found 14% for 2022 and 11% for 2023. Of those who would fail the test, 50% had a temporary contract, 27% an agency contract, and 14% an on-call contract; 6% had a permanent one. Their average monthly wage was €2,701, against €3,954 for those who pass. On 290,225 new benefits in 2025, 12% is roughly 35,000 a year; that arithmetic is ours, not UWV's.

The measure aimed at ageing has been withdrawn. What remains reaches not people in their sixties but flexible workers with an average age of 38.

• The reserve above 60 is not a theory. In 2025, 844,000 people aged 60 to 65 and 412,000 aged 65 to 75 were in work. Anyone who screens out older candidates screens out the fastest-growing group of workers.
• The state pension age stays at 67 through 2027 and moves to 67 years and three months in 2028. Someone who is 62 today has five years left on the market. Plan a career conversation on that horizon, not on 65.
• The stricter entry test, if it comes, reaches temporary workers for half and agency workers for a quarter. Expect flexible workers to weigh what a contract is worth, and a permanent position to carry more persuasive weight.
• Capping the maximum benefit at twelve months would cut the average awarded entitlement from twelve to seven months, and from 21 to 11 months for 55 to 60 year olds. If that measure returns, the pressure to find work fast will be greatest on older job seekers.
• The ratio is stable, its composition is not. The pension share is falling and the benefit share has risen since 2022. Anyone reading the labour market of 2027 should watch the inflow into unemployment and disability benefits, not the ageing curve.

The lead was the CBS release of 8 September 2026, 56 benefit recipients per 100 workers, stable for four years. The figures were read from the custom table beneath it, benefit recipients per 100 workers 2001 to 2025, from the register-based System of Social Statistical Datasets, measured each November. A worker there is a resident whose main source of income is work, with no age or hours threshold; that is a different population from the employed labour force in the labour force survey, 8.51 million against more than 9.9 million, and the two are not mixed here. A pension recipient is a resident aged 55 or over whose main income is the state pension, an occupational pension, a survivor's pension, or an annuity. The years 2024 and 2025 are revised provisional and provisional.

Participation: CBS StatLine 85267NED, labour participation of older people, net participation and employed labour force by age band, annual averages 2015 to 2025. For 2010 the predecessor 82914NED was used, which sits on the survey design used before 2021; in the overlapping years the two series differ by up to one percentage point. State pension age: the government's page on the AOW age per year, and the acts of 2012, 2015, and 2019 that fixed the steps.

Europe: Eurostat lfsa_ergan, employment rate by age, whole population regardless of citizenship, both sexes, 2010 to 2025, for the Netherlands, Belgium, Germany, France, Denmark, Finland, Sweden, Norway, and the EU-27. The table is broken down by citizenship; without that dimension set to the total, the first category comes back, EU citizens, and Denmark drops out from 2021. Every country carries a break in series in 2021 from the new survey regulation; France follows a different definition from 2021. Working life: Eurostat lfsi_dwl_a, 2010 and 2025.

Unemployment benefit: UWV, quantitative information 2025 (benefits in payment at the end of December, average duration), figures and trends January 2026, and the tables to the June 2026 labour market news flash (new benefits per year, stock at the end of June); the end of July 2026 figure comes from the joint CBS and UWV release of 20 August 2026. UWV counts benefits, not people, and the stock includes rights on which nothing is paid. The analysis of the entry test is in the answers of the minister of Social Affairs to questions from the member Patijn, 31 August 2026; UWV calls the result an indication, because payroll returns carry hours per period and those were spread evenly over weeks. The CBS analysis of the 2022 and 2023 cohorts the answer refers to, inflow into unemployment benefits by weeks worked, was not read here.

Policy: the budget table to the coalition agreement of 30 January 2026, rows 56, 57, and 60, and the answers to factual questions of 24 February 2026, tables 24 and 31. The withdrawal of the pension age link and the suspension of the unemployment and disability plans are in the cabinet's letter to the unions FNV, CNV, and VCP of 26 May 2026, copied to parliament. No formal amendment of the budget table existed on 10 September 2026; the budget memorandum of 15 September 2026 is the first document that can settle it. CPB and PBL, analysis of the 2026 to 2030 coalition agreement, February 2026, pages 16 and 17 for the employment effects; that analysis predates the withdrawal and has not been repeated.