INSEE projected a peak of 30.5 million workers in 2040 in a July 2022 document, after which this number will drop to 28.3 million in 2080. Behind this figure lies an intergenerational contract under mounting pressure. The pay-as-you-go pension system rests on three variables—contributions, pensions, retirement age—and adjusting all three simultaneously becomes harder as the base shrinks.
The essentials
- France’s active population will peak at 30.5 million in 2040, then decline to 28.3 million in 2080 (INSEE, July 2022).
- In 2020, the expected duration of retirement was estimated at 23.5 years for men and 27.1 years for women in France; INSEE projects 5.2 million additional people aged 70 or over between 2021 and 2070, compressing both sides of the balance sheet.
- The pay-as-you-go system must arbitrate between three levers: contribution rate, pension level, and retirement age. COR projections show persistent imbalances in most of its scenarios, without establishing absolute impossibility of maintaining current parameters.
- Agnès Verdier-Molinié diagnoses a structural budgetary wall: the contraction of the active base coincides with rising dependency needs, complicating the maintenance of transfer equilibrium.
- The system’s governance remains to be defined: an institutional architecture enabling continuous adjustment would prevent each revision from becoming a political crisis.
The 2040 peak is not just another figure
Thirty years separate France from the moment its active population begins to shrink. INSEE projects 30.5 million workers in 2040. By 2080, this number will be 28.3 million, a decline of 7 percent.
What makes this figure structurally different from past fluctuations is its coincidence with the exit from the labor market of the last cohorts of the baby boom generation. Between 2021 and 2070, INSEE projects a growing number of people aged 70 or over, the point at which dependency becomes statistically likely. France will thus simultaneously finance less and more: fewer workers to contribute, more needs to cover.
The article Population active en recul : quels choix politiques dans les pays de l’OCDE? documents how other OECD countries face comparable trajectories. No single answer emerges, but all credible options involve a reform of the retirement schedule or the structure of contributions.
The duration of retirement varies by gender and generation. The increases in life expectancy recorded since the 1980s have included a significant portion of leisure time after work. This fact, often overlooked in retirement debates, directly affects system arithmetic: stable retirement duration means the burden per worker increases mechanically as the number of workers falls.
The arithmetic of the worker-to-retiree ratio
A pay-as-you-go system works like a pipe. Contributions from workers enter at one end; pensions for retirees exit at the other. The pressure in the pipe depends on the ratio between the two flows.
In France, this ratio has evolved unfavorably since the 1970s. There were roughly four workers for every retiree in the 1960s; this ratio has degraded significantly since, according to the Pension Advisory Council (COR). COR projections, based notably on INSEE demographic assumptions, reveal persistent imbalances after 2040 in most of its scenarios.
The 2023 reform, which raised the legal retirement age to 64, aimed to partly correct this imbalance through the most direct lever: keeping workers in the system longer. Its mechanical effect on accounts is real but limited. The figure of 30.5 million in 2040 does not incorporate the 2023 reform; the post-reform variant projects 31.2 million workers in 2036. The structural demographic problem persists beyond that.
The three variables of the system function as communicating vessels. Increasing contributions weighs on labor costs and workers’ purchasing power. Lowering pensions affects households whose incomes are often modest and fixed. Raising the retirement age runs up against life expectancy inequalities: a blue-collar worker and a senior executive do not reach 64 in the same health condition. The report “Birth Still Determines Destiny, Public Money Reinforces This Rule” documents precisely how life trajectories diverge by social background, which complicates any uniform retirement age increase.
The budgetary wall is demographic, not cyclical
Agnès Verdier-Molinié, director of iFRAP, has argued for several years that France faces a structural budgetary wall. In Face au Mur, she documents how social transfers and local government spending have absorbed budgetary margins to the point of making any consolidation painful in the short term. Her prescription: a deliberate reduction in operating expenses; otherwise the debt spiral becomes self-sustaining.
INSEE projections add a demographic dimension to the fiscal diagnosis. In the INSEE variant incorporating the 2023 reform, the active population reaches a peak of 31.2 million in 2036, at the moment when dependency needs increase sharply. It is the very structure of the baby boom that produces this scissor effect.
What this means concretely: a reduction in the state’s current spending, combined with rising dependency needs, makes adjustments complex. Verdier-Molinié is right that the status quo poses challenges. But the scale of the transformations suggests that multiple combinations of the levers of taxation, pensions, and retirement age can be mobilized. Society will also need to collectively decide what share of the dependency risk remains mutualized, and at what level.
Lessons from European comparisons
France is not an isolated case. Germany, Italy, Spain, and Poland face comparable or more severe demographic trajectories. The 2024 European Commission Ageing Report documents the diversity of policy responses and their differential effectiveness.
Sweden established a notional accounts system as early as 1998: pensions are automatically indexed to demographics and growth, which absorbs shocks without political reform at each cycle. The system adjusts continuously, without crisis. Germany adopted a demographic brake mechanism in 2004, which mechanically ties contribution rates and pension levels to the worker-to-retiree ratio. These devices do not eliminate the pain of adjustment—Swedish pensions actually fell during the 2001 and 2008 crises—but they prevent adjustment from being blocked until it becomes explosive.
Dani Rodrik, an economist specializing in institutions and regulated globalization, offers useful complementary insight here. According to him, what distinguishes systems that hold together from those that collapse is not so much the level of spending as the quality of their governance. Credible automatic rules allow a government to promise sustainability without promising immutability. France, by contrast, tends to legislate rigid parameters, then modify them under political pressure, which produces precisely the repeated crises that automatic-rule systems avoid.
Three scenarios for the decades ahead
After 2040, three major trajectories emerge. None is politically neutral.
The first scenario extends current logic: successive parametric reforms—retirement age, contribution period, pension indexation—absorb shocks over time. This path is workable but politically costly. Each reform is experienced as a breach of contract, fueling distrust of institutions. France has already traveled this path several times since 1993; the results in terms of legitimacy are poor.
The second scenario opts for systemic reform toward a points or notional accounts model, which the Delevoye report had sketched before its abandonment in 2020. This model makes trade-offs transparent: everyone sees what they contribute and what they will accumulate. It transfers part of the demographic risk to individuals, raising a social justice question that demographic data makes even more acute: blue-collar workers, whose healthy life expectancy is significantly lower than that of senior executives, bear more of the risk of contributing for long in return for little.
The third scenario banks on immigration to broaden the active base. The OECD documents that migration flows can partially correct a demographic deficit, provided integration policies are effective and offered jobs match qualifications. Depending on migration assumptions chosen, the level of active population in 2070 varies significantly: the high migration balance scenario projects approximately 2.1 million additional workers compared to the central scenario. This lever remains partial and does not resolve the question of baby boomer dependency, which is both dated and certain.
These three scenarios are not mutually exclusive. The most likely trajectory is a combination of the three, weighted differently depending on successive political majorities.
An institutional architecture to adjust without crisis
The question posed by the 2040 peak goes beyond pension system parameters. It touches on the governance of the intergenerational contract itself.
The Swedish model and the German mechanism have a point in common: they separate the principle decision, concerning the level of mutualization, from the technical decision, concerning parameter adjustment. The first remains political and democratic; the second becomes automatic, transparent, and predictable. This separation reduces the political cost of each adjustment, perceived as the application of a known rule rather than an arbitrary decision.
In France, this separation does not exist. Parliament votes on specific parameters; each modification requires a vote; each vote becomes a tug-of-war. The COR produces rigorous diagnoses that successive governments use selectively. The result is a series of reforms perceived as arbitrary, even when objectively necessary.
A more robust institutional architecture could operate through an automatic mechanism framed by law: if the worker-to-retiree ratio crosses a defined threshold, parameter adjustments trigger according to a publicly known rule. Parliament retains the capacity to suspend or modify the rule, but must do so explicitly. This type of device transforms a political conflict into transparent technical decision.
The life expectancy and working condition inequalities, which data on social mobility show as persistent, constitute a legitimate argument against any mechanical retirement age increase. A transparent rule offers a stable basis for negotiation, which current legislative instability does not permit.
The 2040 peak is a known, dated, and quantified constraint. Countries that have managed to anticipate it have built institutions capable of adjusting continuously, rather than deferring decisions until they become urgent.
Sources
- INSEE, Projections de population active 2021-2070, juin 2026
- Agnès Verdier-Molinié, Face au Mur, iFRAP
- European Commission, The 2024 Ageing Report, Directorate-General for Economic and Financial Affairs (no link: report available on the European Commission website)
- Pension Advisory Council (COR), annual reports 2023-2025 (no link: available on cor-retraites.fr)
- OECD, Long-term Growth Scenarios, Department of Economics (no link: available on oecd.org)
