Pay-as-you-go pensions: how the French system works, and why “42 schemes”?

“We must put an end to the 42 special schemes.” The line ran on a loop for years. Yet it packs two errors into a handful of words: there aren’t 42 special schemes, and the system isn’t just a pile of privileges to trim. Before judging the French pension system, you have to know how it’s built. That’s what this article is for — the first in a series that will then open each major scheme, one by one, before comparing France with its neighbours.

Let’s start at the beginning: what exactly is “pay-as-you-go”?

Pay-as-you-go: paying today’s retirees, not yourself

The French system rests, for more than 95%, on pay-as-you-go. The principle fits in one sentence: the contributions levied this month on workers’ wages are used, this month, to pay retirees’ pensions. There is no pot in your name sleeping somewhere: your contribution immediately funds someone else’s pension, and your own future rights will be paid by tomorrow’s workers. It’s a contract between generations, not a savings account.

The other model, funded pensions, works the opposite way: each person saves and invests for their own retirement, and recovers, decades later, the accumulated capital plus interest. In France, funded pensions exist — the PER, workplace retirement savings — but they remain optional and marginal, sitting on top of the mandatory base.

The sensitive point: demographics

Because pay-as-you-go has workers paying for retirees, its balance depends on a single ratio: the number of contributors per retiree. And that ratio is eroding. It was around 2.1 contributors per retiree in 2000; it had fallen to ~1.7 by 2023¹ (30.4 million contributors for 17.2 million retirees). Fewer workers funding more pensions: that’s the whole stake of successive reforms — and the reason we’ll come back to it at the end of this article.

The real architecture: base + complementary, for everyone

Nobody has “a” pension: everyone has at least two tiers, both mandatory and pay-as-you-go.

1. A base scheme. For the private sector it’s the general scheme, run by CNAV (“l’Assurance retraite”); for civil servants, the SRE (State) or CNRACL (hospital and local government); for others, a special scheme or a dedicated fund.
2. A complementary pension scheme. For the private sector that’s Agirc-Arrco, run on a points basis: contributions buy points, converted into a pension at retirement.

On top of these two mandatory tiers comes, optionally, a third: funded savings (the PER, workplace retirement plans) — the subject of a later article in this series.

The general scheme’s pension is calculated on the reference average salary (the 25 best years), to which a rate and any pension discount / premium apply depending on the number of validated contribution quarters. But each scheme keeps its own rules: the civil service calculates on the last six months of pay, and each special pension scheme adds its own credits. That diversity of funds is what feeds the famous “42”.

Here are the main schemes (non-exhaustive):

“42 schemes”: what the figure really means

The number comes up in every debate: “there are 42 pension schemes.” It’s accurate — but almost always mis-stated. People hear it as “42 special schemes,” implying 42 pockets of privilege. That’s wrong. The correct wording is “42 mandatory pension schemes”²: it counts every distinct affiliation situation once base and complementary are combined — to the point that Agirc and Arrco, or the liberal professions’ base fund and its professional sections, are counted separately.

The genuine special schemes — those with their own calculation rules and their own fund — number about ten, not forty-two². The nuance isn’t a detail: it’s the difference between “the system is a jumble of 42 privileges” and “the system is a historical assembly of occupational schemes, about ten of them special.”

2023: a reform… then its suspension in 2026

The last major reform — the law of 14 April 2023³ — did two things. It pushed the legal age from 62 to 64 (3 months per birth cohort from those born in September 1961, reaching 64 for the 1968 cohort) and sped up the lengthening of the contribution period toward 43 years / 172 quarters from the 1965 cohort. It also closed several special schemes to new entrants on 1 September 2023 — RATP, IEG, notaries’ clerks, Banque de France, the CESE⁴ — under the grandfather clause: staff already in place keep their scheme, only new entrants move to the general scheme. SNCF, for its part, had already been closed since 1 January 2020⁵.

Then, in late 2025, the calendar reversed. The 2026 social-security financing act suspended the phase-in⁶: from 1 September 2026 to 1 January 2028, the age stays frozen at 62 years and 9 months and the duration at 170 quarters, instead of climbing toward 64 and 172.

What the arithmetic says

Pushing back the age is the most powerful lever a pay-as-you-go system has, because it acts twice in the same direction: a person who works one more quarter becomes an extra contributor (more revenue) and one fewer retiree that quarter (less spending). That’s also why suspending the measure quickly costs money: about €0.4bn in 2026, €1.8bn in 2027, and a cumulative deficit estimated at +€18–20bn by 2035 if the suspension lasted⁷.

Above all — and this is the key point — the 2023 reform was already insufficient. Even with the move to 64, the COR projects the system in deficit by around €15bn in 2035 and €30bn in 2045⁷. The 2023 debate was therefore not about “balance versus deficit,” but “slightly less deficit versus slightly more.” Backing away from a measure that already wasn’t enough mechanically moves the system further from balance.

Does that make it an “error”? On the direction, the arithmetic leaves no room for doubt: with contributions and pensions unchanged, freezing the age widens the gap. On the magnitude, it’s more nuanced. The “double effect” is a theoretical maximum: some of the seniors kept in work on paper are actually unemployed or on disability, and don’t become contributors — the OFCE points out that the net gain is smaller than the gross one⁸. And “error” remains, ultimately, a value judgement: one can deliberately prioritise arduous-work fairness, senior employment or equity at the cost of a larger deficit. What one cannot do is pretend that freezing the age is neutral for the books.

Key takeaways

• France runs on pay-as-you-go: workers pay retirees, with no individual pot; the balance depends on the contributors-to-retirees ratio, now down to ~1.7.
• Every worker has at least two mandatory tiers (base + complementary); funded savings are only an optional top-up.
• The “42 schemes” are 42 mandatory schemes (base × complementary combinations), not 42 special ones — those number about ten, several closed to new entrants since 2020–2023.
• The 2023 reform (age 64) has been suspended until 2028; it was, in any case, insufficient to balance the system, which the COR sees in deficit even with it.

Next episode: we step into the first special scheme, that of the SNCF railways — its history, its rules, and what “closed in 2020” really means.

This article is general information and does not constitute advice. The rules and figures cited are those known in mid-2026 and change with each budget and social-security financing act.