A gradual expansion from the post-war period to the 1980s

1 The French pension system’s foundations were laid in the immediate post-war period, but its development was gradual. The earliest figures given by social security accounts date back to 1959. In that year, when France was much less affluent than it is today, expenditures on old-age and survivors’ benefits represented only 5.2% of the gross domestic product (GDP) [1], the retirement age was 65, and the system guaranteed a far from comfortable living standard for pensioners. By 1970, their standard of living still represented only 70% of the standard of living of people of working ages [2].

2 The system acquired its current dimensions throughout the 1970s and the early 1980s, with the adoption of more generous rules for the basic regimes and the extension of coverage by the complementary regimes. This led to a gradual alignment of the mean living standard of retiree households with that of the population as a whole. Today, taking account of the smaller size of these households, it is even slightly higher. These improvements occurred in parallel with a decrease in the age of withdrawal from the labour market, thanks first to the development of early retirement schemes and then, in 1984, to the lowering of the full-rate retirement age to 60 for those who contributed for at least 37.5 years. At the time, this condition was fulfilled by the vast majority of the population, at least for male workers—less so for women.

3 Pensions had therefore become much higher and paid out for much longer periods because, in addition, life expectancy had started increasing again after having stagnated in the 1960s. To cover the additional costs, the share of GDP devoted to spending on old-age and survivors’ benefits was raised substantially, reaching 11% in the mid-1980s.

4 It is against this backdrop that doubts about the system’s sustainability gradually emerged, with demographic projections that had started exploring perspectives for after the year 2000. The future they envisioned for 2040 was a doubling of the number of over-60s with respect to that of the population aged  0–59, and its easily stylized consequences: either a near doubling of the share of GDP devoted to pensions or a halving of pensioners’ relative income level. The possible third option was to compensate population ageing by a very substantial shift of the frontier age between work and retirement, up to about age 70. Given the enormity of all these figures, it seemed impossible to have the problem solved by acting on just one of these levers. A solution combining all three was a much more realistic option. The reforms implemented up to now have indeed done so, but not without debate and hesitation.

Drivers of population ageing: the importance of accurate diagnosis

5 Early debate on pensions was distorted by a misunderstanding of the drivers of population ageing. There was a strong French tradition of concern for the risk of population decline, inherited from the interwar period. The culprit for pension problems thus appeared obvious: the below-replacement fertility rate, the impact of increasing life expectancy being left aside.

6 What was wrong with this diagnosis? It is true that fertility had been declining since the mid-1960s and that generation replacement was no longer totally assured. But could the pension problem be resolved simply by re-increasing birth rates? To do so would have demanded a spectacular increase, to levels higher than those reached during the baby boom, or, alternatively, a totally unrealistic mass inflow of migrants. In a population where lengthening life expectancy leads to ageing ‘from the top’, the only way to counter this trend is a rejuvenation at the base or in the middle of the age pyramid, with the rapid population growth that this implies. The baby boom had achieved this for several decades, but such a growth phase had to come to an end. It did so in 2006, when the first baby-boom cohorts reached age 60, marking the onset of a period of rapid ageing that is still ongoing today [3].

Illustration

Illustration

7 This initial focus on fertility did not help to advance the debate. It long fuelled doubts about the reality of the problem. With a problem described as being due only to below-replacement fertility, each slight upturn in the fertility rate was unduly interpreted as a sign that reforms might be avoidable. Moreover, the idea of a pension deficit due to a lack of contributors prompted the idea of saving the system by a return to funded pensions, with capital paying the costs in place of workers’ contributions. Yet faced with the challenge of increasing life expectancy, funded pensions are not intrinsically any more robust than a pay-as-you-go system. Longer periods of retirement necessarily cost more, whatever the mode of funding.

1993, 2003, 2010, and 2014: the four main reforms

8 This controversy between advocates of pay-as-you-go and funded systems dominated the debate for years and was still doing so in 1993, the year of the first major reform aiming at containing pension expenditures. Its two emblematic measures were the lengthening of the contribution period required for a full pension, with a planned increase from 37.5 to 40 years between the 1933 and 1943 birth cohorts, and a lengthening of the period used to calculate the reference wage serving as a basis for the replacement rate, from the 10 best years to the 25 best years.

9 More discreetly, this reform also confirmed a practice introduced in the late 1980s, namely the transition from a ‘wage’ reference to a ‘price’ reference for two central components of pension entitlement calculations: the revaluation of past wages used to calculate the first pension and the indexation of the pension after it has been awarded.

10 The consequences of this dual ‘price’ reference merit close examination. The average of the 25 best annual wages revalued in line with past inflation is inevitably lower than that of the same wages revalued in line with past average wage growth. Moreover, current pensions increase more slowly if they are indexed on prices than if the fruits of growth are fully shared between the active population and retirees. Overall, under the current medium scenario of the French Pensions Advisory Council (Conseil d’orientation des retraites [COR]), retirees’ living standards should start falling again with respect to the working-age population, by around 20% over the long term, returning to the 1980s level of around 80%–90%.

11 Subsequent reforms of the basic regimes made no further changes in this area. As the driving role of life expectancy finally gained recognition, their focus shifted to the retirement age. The message that ‘there will be no one to pay for our pensions tomorrow’ had been progressively replaced by the need to manage the fact that, ‘every year, we were gaining 3 months of life expectancy’, which was the rate at which longevity was increasing.

12 One easily sees the window of opportunity that has been opened by such a shift. If life expectancy increases by 3 months each year, retirement age can be raised without necessarily shortening the time spent in retirement. For example, 2 of these 3 months can be allocated to working and the rest to retirement in order to preserve a two-thirds/one-third ratio between the duration of working life and that of retirement. This is what the 2003 reform sought to do by further increasing the contribution period required for a full pension. The 2010 reform followed similar lines, raising the minimum retirement age from 60 to 62. That of 2014 again increased the contribution period, raising it to 43 years for the 1970 birth cohort.

13 But this policy of sharing life-expectancy gains has its limits. It addresses the gains achieved after its implementation, but not the future consequences of past mortality declines or the knock-on effects of the baby boom. To deal with these components of ageing, reforms have had to rely on the other two instruments: the relative decrease in entitlements stemming from indexation on prices and an increasing share of GDP devoted to pension expenditures.

The trajectory of the ratio of pensions to GDP

14 How big of an increase? As mentioned earlier, the ratio of pensions to GDP had already risen to around 11% in the 1980s, creeping up to 12% by the early 2000s. In its first report released in 2001 [4], the COR projected this ratio under the scenario of a 1.6% increase in productivity, allowing price indexation to play a substantial role in balancing the system. This projection had numerous variants, including one with an over-indexation equal to prices plus 1.2% per year, thus approaching the previous rule of indexation on wages.

15 This variant confirmed the order of magnitude cited above, i.e. a doubling of the share of pension spending in the GDP in the absence of any other form of adjustment, with a ratio of 20% reached in 2040 (Figure). The reference scenario showed how the 1993 reform had already brought this predicted ratio down to 15.7% of GDP. The 2003 reform further flattened the trajectory. But its impact was affected by an unanticipated event, the economic crisis of 2008–2009 and its aftermath. French GDP growth failed to meet its forecast levels. It is this slowdown that motivated the reforms of 2010 and 2014.

16 What happens in a case like this is easy to understand. Overall pension spending continues steadily along the same trajectory while the resources that fund it shrink or stagnate. The effect is quite significant. Today’s GDP is around 15 points lower than the level projected before the crisis. For a ratio of pensions to GDP of around 12%, a denominator 15 points lower raises the ratio by 1.8 points. Combined with the effects of entry into retirement of the large baby-boom cohorts, this factor has raised pension spending to the new level of around 14% of GDP.

What now?

17 On this basis, what is the outlook for the future? With the cumulated effects of pre-crisis reforms, the ‘crisis’ reform of 2010 and the post-crisis reform of 2014, pension spending should, according to the COR, remain stable at around 14% of GDP over the long term in the low but plausible scenario of annual productivity increases of just 1% [2].

18 If this scenario were confirmed, we could say the pension problem has been resolved by applying in roughly equal measure the three balancing options available in the 1990s. Of the 9-point increase in the ratio of pensions to GDP that would have occurred without any reform, a rise of around 3 points has already become a reality, due partly to the crisis. The remaining 6 points would be avoided thanks to changes in the retirement age and retirees’ relative living standard: a progressive increase to around age 64—rather than 70—for the effective retirement age and a progressive reduction of around 20%—as opposed to a halving—of retirees’ relative purchasing power.

19 This adjustment of entitlements is what is required if one wishes to avoid any further increase in the ratio of pensions to GDP. The stabilization of this ratio is an objective that can be debated, of course, but without forgetting the many other pressing economic, social, and environmental issues to be addressed. One can also underline that this apparent regression in entitlements by no means signifies that the sums paid in by current contributors will be lost for them. A pay-as-you-go system whose size is constant with respect to GDP offers, by construction, an average yield equal to the rate of economic growth. It would thus remain positive, comparing favourably with the current yields of certain financial investments. Admittedly, retirees would return to the relative living standard of the 1980s and 1990s, but with pensions at higher absolute levels and paid out for longer than at that time. This relative living standard for retirees is that already observed in many other countries also confronted by population ageing.

20 This overall balance, however, is highly sensitive to future growth assumptions. With indexing on prices, the difference between pensions and wages widens all the more with rapid growth in productivity, as it is productivity gains that allow wages to rise faster than prices. If gains topped 1% per year, retirees’ purchasing power would drop more sharply, and the share of pensions in GDP would therefore decrease. Under the COR’s high scenario, this share would even return to the pre-crisis level (Figure). But if productivity growth were to fall below the level of 1% used in the low scenario, as might well be the case, everyone would suffer in the end, but retirees less rapidly and slightly less severely than others: their relative living standard would improve, and the share of pensions in GDP would start increasing again. Not to mention the potential impact of another recession.

21 This sensitivity to productivity assumptions and the impact of the 2008–2009 crisis are two aspects of the same problem, that of how adjustments to essentially unpredictable changes in economic growth should be shared between the active and retired populations. This is a problem of navigation, of which demographic uncertainty is another aspect. While ageing is inevitable, its exact intensity will depend on combined trends in life expectancy, fertility, and migration. Some of the recent changes in these parameters work in favour of faster ageing, while others push in the opposite direction; as yet, we do not know which factors will predominate.

22 In short, past reforms have placed pension expenditures on a course that should remain stable through steady and moderate winds. In this sense, we can say that balance has been restored. The problem will be to guarantee this balance if the winds turn or change in intensity and, faced with these uncertainties, to ensure that current and future retirees trust that everyone is sailing in the same ship, or at least in the same type of ship. For what previous reforms have failed to address is the perception of unequal treatment between different population categories. The problem arises both for calculations of own contributory entitlements, which remain very heterogeneous, and for non-contributory or derived entitlements, such as survivors’ pensions. It is natural for a pension system not to treat everyone identically, but inequalities of treatment need to follow well-defined rules that clearly target those for whom they are most justified.