1
P
ENSION PROJECTIONS FOR THE
2021
A
GEING
R
EPORT
C
OUNTRY FICHE
F
RANCE
Final Version – March 2021
2
Table of contents
1. Overview of the pension system .................................................................................. 3
1.1. Description ........................................................................................................... 3
1.2. Recent reforms of the pension system included in the projections .................... 11
1.3. Description of the actual ‘constant policy’ assumptions used in the projection 15
2. Overview of the demographic and labour forces projections .................................... 16
2.1. Demographic development ................................................................................ 16
2.2. Labour force ....................................................................................................... 17
3.1. Extent of the coverage of the pension schemes in the projections ..................... 20
3.2. Overview of projection results ........................................................................... 21
3.3. Description of main driving forces behind the projection results and their
implications .......................................................................................................................... 25
3.4. Financing of the pension system ........................................................................ 35
3.5. Sensitivity analysis ............................................................................................. 37
3.6. Description of the changes in comparison with the 2006, 2009, 2012, 2015 and
2018 projections ................................................................................................................... 39
4. Description of the pension projection model and its base data ................................. 44
4.1. Institutional context in which the projections are made ..................................... 44
4.2. Data used to run the model ................................................................................. 44
4.3. General description of the model ....................................................................... 45
4.4. Assumptions and methodologies applied ........................................................... 46
4.5. Additional features of the projection model ....................................................... 47
5. Appendix ................................................................................................................... 48
A. Methodological annex ............................................................................................ 48
B. Retirement ages of the French pension system ...................................................... 53
C. Full pension contribution period ............................................................................ 54
E. Projected and disaggregated new public pension expenditure (old-age and early
earnings-related pensions) .................................................................................................... 58
F. Overview of the sensitivity tests ............................................................................ 60
G. Panorama of the main pension schemes ................................................................ 62
6. References ................................................................................................................. 64
3
1. Overview of the pension system
1.1. Description
- A system made of different schemes for old-age and survivor pensions
The French pension system is essentially a pay-as-you-go system financed by contributions from both
workers and employers. The description of the pension system and the projections are based on the
legislation in force in 2020.
The French pension system is based on several schemes depending on the professional sector or
occupational status. The private sector employees scheme (CNAV) is the largest one. These schemes
follow different rules. All workers are affiliated, according to their profession, at the same time to a
basic and a mandatory complementary scheme. They can belong to several basic schemes during their
career: they then receive several pensions at retirement.
Table 1.1 - Outline of the French pension system
4
Note: Regarding acronym developments, please refer to the glossary.
*Self-employed workers (artisans and retailers), affiliated to the Social Regime for the Self-Employed (Régime Social des Indépendants - RSI) until December 31, 2017, are affiliated to the Social
Security for the Self-Employed (Sécurité Sociale des Indépendants - SSI) from January 1, 2018, whose management is gradually assumed by the French National Old Age Insurance Fund (Caisse
Nationale d'Assurance Vieillesse - CNAV).
5
Special Schemes
Within the special pension schemes, there are three families:
• The special civil service scheme (civil and military servants) ;
• The regime for companies and public establishments ;
• Other schemes grouped around a profession or company.
The main special schemes include :
• The scheme for civil servants, magistrates and military personnel covered by the Civil and Mil-
itary Retirement Pensions Code ;
• The SNCF's scheme with the SNCF provident and pension fund ;
• The scheme for Clerks and Notary Employees with the Pension and Welfare Fund for Clerks
and Notary Employees ;
• The RATP's scheme with the RATP staff pension fund ;
• The Seamen's regime with the National Establishment for the Invalids of the Navy ;
• The EDF-GDF Electricity and Gas Industries scheme with the National Fund for Electricity and
Gas Industries ;
• The Cult Plan with the Old Age and Illness Insurance Fund for Religious Leaders ;
• The Senate's parliamentary system with the Senate's autonomous social security fund, which
has not been aligned with the civil service system ;
• The parliamentary regime of the National Assembly with the Social Security Fund of the Na-
tional Assembly, aligned since January 2018 with the Civil Service regime ;
• The regime of the French Comedy ;
• The regime of the Autonomous Port of Strasbourg ;
• The Banque de France regime.
Special schemes represent about 4.5 million pensioners, i.e. 25.7% of French pensioners and 4.7 million
contributors, i.e. 3.4% of the labour force.
Special schemes workforce Contributors pensioners
Civil and military servants
2 058 000 2 355 000
Territorial and medical civil servants
2 223 000 1 155 000
Sate-owned industrial civil servants
35 772 103 682
Electricity and gas industries
146 103 164 895
SNCF
142 943 261 033
RATP
42 483 44 316
Clerks and Notary Employees
47 618 73 090
Seamen's
20 045 117 830
Banque de France
12 029 14 891
The mining regime
3 401 303 970
Paris National Opera
1 825 1 999
French Comedy
350 401
6
Retirement age
The retirement age depends on the behaviour of the new pensioners. There is a legal minimum age
1
and
incentives to retire later.
- People can retire when they reach the earliest retirement age (62 for the 1955 generation and
the following ones), with a penalty if they do not meet the required contribution period condition
(43 years from the 1973 generation onward).
- They can also delay their entry into retirement in order to obtain a full pension which is granted
for people with the required contribution period or above the statutory retirement age (also
called full pension age, 65 up to the 1951 generation, 67 for generations born in 1955 and after).
People who are allowed to retire with a full pension (as they meet the age and contribution
period conditions) but who decide to keep working will receive a bonus on their pension pro-
portional to the number of extra years worked.
There are some exceptions to the legal retirement age. The most important one is dedicated to people
who have started working at a very young age and have validated more than the required time (see
details infra). In the public sector, for some special branches among “active service” (policemen, nurses,
etc.), the minimum retirement age is 57 years old
2
. In general, there is no gender difference in the
eligibility requirements.
Table 1.2 – Qualifying conditions for retirement
20
20
20
3
0
20
4
0
20
5
0
20
6
0
20
7
0
20 contribution years *
statutory retirement age ** 67,0 67,0 67,0 67,0 67,0 67,0
earliest retirement age 62,0 62,0 62,0 62,0 62,0 62,0
penalty in case of earliest retirement age
25,0% 25,0% 25,0% 25,0% 25,0%
25,0%
bonus in case of late retirement *** 0,0% 0,0% 0,0% 0,0% 0,0% 0,0%
40 contribution years *
statutory retirement age ** 63,8 64,5 65,0 65,0 65,0 65,0
earliest retirement age 62,0 62,0 62,0 62,0 62,0 62,0
penalty in case of earliest retirement age
8,8% 12,5% 15,0% 15,0% 15,0%
15,0%
bonus in case of late retirement *** 16,3% 12,5% 10,0% 10,0% 10,0%
10,0%
* We assume that people have accumulated 20 or 40 years of contribution at their earliest retirement age the given year.
Statutory retirement age is then reached after that year.
** Statutory retirement age is reached when full pension is attain via contribution period or legal retirement age
*** We assume late retirement to be the legal automatic full rate pension age (67 yo. in 2020 and after)
NB: we do not distinguish between women and men since they face the same legislation.
Benefit fomula
a. Old age pension
The rules to calculate pensions differ from one scheme to another. We present here only the formula
used to calculate the two components of the pension in the private sector (basic pension from the CNAV
and complementary pension from the AGIRC-ARRCO) and in the public sector.
1
Rules may differ from the general situation in certain schemes, for instance, the complementary scheme of
independent professions.
2
Since the 2014 reform, the minimum retirement age for “active service” is increased from 55 years old for generation 1956
and before, to 57 years old for generation 1960 and after.
7
Basic private sector pensions (CNAV and aligned schemes)
In the basic private sector (CNAV) and the aligned schemes (RSI and MSA workers), the pension P is
calculated according to the following formula:
P=ref.wage ×Min (1,
D
T
)×t
Three factors compose that formula:
The reference wage is the average wage over the 25 best wages (up to the social security
ceiling, € 3 428 per month in 2020), with past earnings valorized in line with price inflation.
The coefficient of proratisation Min (1,D/T) with D being the contribution period, that is the
number of years validated by the insured and T, the reference length. In other words, the
pension is reduced in due proportion whenever D < T. For people born in 1959 (who will be
62 in 2021), T equals 41.75 years, this value will increase up to 43 years for people born in
1973 or after.
The pension rate t. The standard rate is 50%.
However, in order to foster senior participation rate in the labor market, either a penalty or a bonus can
be applied under certain conditions:
• A penalty is applied to the pension rate when the pension is withdrawn before the full pension age
if the contribution period is lower than the reference one (D < T). The deduction is then calculated
as Min [Full pension age - Age, (T-D)] multiplied by the rate of deduction (1.25% per missing
quarter from the 1953 cohort onward). The new pension rate t’ is given by:
= × (1 − 1.25% × number of missing quarters)
• Conversely, the pension is augmented by a premium when individuals decided to continue to work
although they had met the conditions for a full pension. The premium is calculated as
Min [Age - Minimum retirement age, (D-T)] multiplied by the premium rate (1.25% per quarter).
The new pension P’ is given by:
′ = × (1 + 1.25% × number of extra quarters worked)
There is a minimum contributory pension (named minimum contributif) for individuals who meet the
requirements for a full pension (ie. they are 67 years old or they have contributed long enough for being
granted a full pension before the age of 67) amounting to € 642,93
3
per month in January 2020. This
minimum is price-indexed.
Mandatory complementary pension (AGIRC-ARRCO for all private sector workers)
Complementary schemes for private sector employees are pay-as-you-go point systems that serve
defined-contribution pensions. Contributors acquire each year a certain number of points through their
own contributions and those of their employer, calculated on the basis of an acquisition rate τ
t
applied
to a part of their gross wage. The acquisition rate τ
t
equals the contribution rate of the scheme divided
by 127%. The purchase price of each point, called “reference wage”, depends on the year considered.
In January 2020, it amounted to € 17,3982.
Number of points acquired in year t = τ
t
× (Gross wage
t
/Purchase price of a point
t
)
3
Only individuals whose total pension (basic + complementary) does not exceed € 1 191,57 per month (in 2020)
are entitled to the minimum contributory pension of the basic scheme. A higher minimum contributif also exists
for people having contributed 120 quarters.
8
At retirement, the transformation of accumulated points into a pension benefit is a function of the
contributor's age, the contribution length and the selling price of a point at that date. Complementary
pension is then calculated as follows:
Pension = Total number of points acquired × Value of a point × Shortfall coefficient
“Full” complementary pension is granted only to those who qualify for a full basic scheme pension. In
case one retires before fulfilling the requirements for a full pension as defined by the CNAV, the value
of the point is adjusted downwards by means of a “shortfall coefficient” (cf. Table 2.2). For example, a
pensioner with 4, 12 or 20 contribution quarters missing will see his pension pro-rated by 0.96, 0.88 or
0.78 respectively.
Following the 2015 agreement (cf. 1.2), a new system of “solidarity coefficients” and “increase
coefficients” is factored into the calculation of the complementary pension benefit and is based on the
age at which employees acquire full rights to the basic pension under the CNAV general scheme will be
put in place, starting with the generation 1957. The coefficients work in the following way:
• For individuals who retire less than one calendar year after the age at which they are entitled to
a full basic pension, the AGIRC and ARRCO complementary pension benefits are reduced by
a solidarity coefficient of 10% for three years or until they turn 67.
• Individuals who retire between one and two years after that age receive their full pension, with
no solidarity coefficient or increase coefficient.
• For each additional year that the individual delays retirement, the AGIRC and ARRCO comple-
mentary pension is increased for one year by an increase coefficient of 10% (up to a maximum
30%).
• Pensioners exempted from the “general social security contribution” (Contribution Sociale Gé-
néralisée: CSG) and certain precarious categories of pensioners
4
are exempted from the solidar-
ity coefficient (but are subject to the increase coefficient). Pensioners paying the CSG at the
reduced rate
5
are subject to the solidarity coefficient but with a 5% reduction instead of 10%.
Pension in the public service scheme (FPE)
The calculation of the basic pension for public sector workers is very similar to the one in the CNAV:
P=ref.wage×Min (1,
D
T
)×t
Nevertheless the parameters differ from those of the general scheme in two essential aspects:
• The reference wage taken into account is the wage received the last 6 months (excluding
bonuses and other emoluments), as opposed to the average of the best 25 years' wages (including
bonuses) in the private sector.
• The pension rate t is 75%. The 2003 reform introduced also a penalty scheme and a premium
rate, similar to the ones existing for private sector employees.
As in the main basic scheme, the duration T taken into account in the pro rata coefficient is 41.75 years
for people born in 1959 (aged 62 in 2021) and will increase up to 43 years for people born in 1973 and
after.
4
The 30 October 2015 agreement lists the conditions for exemption from solidarity coefficients.
5
The standard CSG rate for pensioners is 6.6%. The reduced CSG rate (3.8%) and exemption from CSG are
subject to means testing.
9
Unlike private sector employees, public sector employees did not receive complementary pensions until
recently. This is why their basic scheme replacement rate is higher. A complementary pension scheme
(RAFP) was introduced in 2005 by the 2003 reform. It is a point system whose contributions are only
based on bonuses, within the limit of 20% of total wage. This scheme provides pensions which are much
lower than those of the private sector complementary schemes.
For pensioners who meet the requirements for a full pension, an earnings-related minimum pension is
guaranteed (called minimum garanti). In 2020, its value was € 14 092,08 per year (€ 1 174,34 monthly)
for a 40 year long career.
b. Survivor pension
The surviving spouse's allowances include the survivor's pension (pension de réversion), which corre-
sponds to a proportion of the pension which the deceased person could have received and which it is
paid to the surviving spouse or ex-spouse and the widowhood allowance (allocation de veuvage), which
is paid to the surviving spouse depending on her income if she does not meet the conditions of age in
order to benefit from a survivor's pension. The survivor's pension
may be granted, even if the deceased
spouse died before retirement or before reaching the minimum retirement age.
Basic private sector survivor pensions (CNAV and aligned schemes)
To benefit from the survivor pension, the surviving spouse or ex-spouse:
• must have been a married to the deceased beneficiary (the Pacs civil partnership and cohabita-
tion do not entitle the partner to the survivor's pension);
• must be at least 55 years old, this age being reduced in certain cases (death or disappearance
before 2009);
• must earn gross annual income that does not exceed €21,112 if she lives alone, or €33,779.20 if
she lives with a partner.
If the deceased spouse has been married several times, the survivor pension is shared between the sur-
viving spouse and the divorced ex-spouse(s). This division is proportional to the duration of each mar-
riage.
The survivors pension is equal to 54% of the deceased spouse or ex-spouse’s basic pension (or the basic
pension she could have benefited from). If the spouse or ex-spouse made 15 years of contributions to
the general pension scheme, the minimum amount of your survivor's pension is €3,478.46 per year in
2020. If the deceased had contributed for less than 15 years, this minimum total is reduced proportion-
ally. In 2020, the amount of the survivor's pension cannot exceed €11,106.72 per year.
Pension in the public service scheme (FPE)
To benefit from the survivor pension, the surviving spouse or ex-spouse of a public servant must qualify
for at least one of the following conditions:
• to have been married to the deceased beneficiary at least four years
• to have been married to the deceased beneficiary at least two years before the deceased spouse
retired;
• to have at least one child with the deceased beneficiary
To benefit from the survivor pension, the surviving spouse must not live with a partner.
10
The survivor pension is equal to 50% of the deceased spouse or ex-spouse’ pension (or the pension she
could have benefited from). It can be increased by half of the majoration the deceased spouse benefited
for having raised at least three children.
c. Non earnings-related minimum pension
People aged 65 and above (or 62 in case of incapacity or disability) whose revenues (including pension
or not) are below a certain ceiling (€ 10 838,40 a year for a single person and € 16 826,64 for a couple
in 2020) are eligible to a minimum pension, named ASPA (standing for “Allocation de solidarité aux
personnes âgées”) that tops revenues up to the aforementioned ceiling. ASPA-related expenditures
amounted to € 3,1 billion in 2018, which represents 1 % of the total amount of pension expenditures.
d. Disability pension
Disability pensions provide a replacement income for people who are completely or partially,
temporarily or permanently, unable to work. These pensions are paid by the public health insurance
schemes.
There are two different earnings-related disability pensions: the “rente Accident du Travail et Maladie
Professionnelle (ATMP)” which is granted when the disability is related to work and the “Pension
d’Invalidité (PI)” which is granted when it is not work-related. When disabled persons receiving a PI
reach the legal retirement age, they become eligible to a full pension replacing their disability pension:
their pension is no longer included in disability expenditures and is transferred into old-age
expenditures’ accounts. This is also true for PI for public servants even if in practice they receive an
only pension before and after legal retirement age. On the contrary, the ATMP pension is perceived is
cumulative with an old-age pension.
These earnings-related disability pensions are a fraction of a reference wage (the average of the ten best
wages for PI and twelve last months for ATMP), depending on the disability level as exhibited in the
following table. It cannot exceed a maximum nor be inferior to a minimum amount.
• for PI calculation, the percentage applied to the reference wage varies according to the disability
class (1st class, 2
nd
class or 3rd class), from 30% (1st class) to 50% class (2
nd
class, with an
additional 40% bonus for a third party for 3rd class); minimum and maximum level also vary
from respectively 292.80€ (1st class and 2
nd
class) per month to 1,418.09€ (3rd class) and from
1,028.40€ (1st class) to 1,714.00€ (2
nd
class) to 2,839.29€ (3rd class)
6
.
• concerning ATMP, the annuel pension (P) equals : P = T x R, with :
= 0.5 × if disability rate ≤ 50%
= 1.5 × – 50% if disability rate > 50%
= ref.wage if ref.wage ≤ °
= ° +
.°
if < ref.wage ≤4°
with R° = € 37,262.59. Revenues above 4R° are not factored into the calculation.
A non earnings-related minimum disability pension (“Allocation aux adultes handicapés” - AAH) is
also considered in the projections : AAH tops revenue of all disabled people up to a ceiling of € 902.70
per month in 2020. Howener, in the absence of a detailed breakdown of recipients and expenditures by
profile, expenses related to Allocation spécifique invalidité (ASI), another non earnings-related
minimum disability pension which accounts for less than 250 M€ in 2018, are not included.
6
Source : Public administration official website : www.service-public.fr.
11
Indexation
All basic schemes pensions, as well as disability pensions (PI, ATMP), are price-indexed. Past wages
taken into account for the pension calculation are also revalued with the Consumer Price Index
(excluding tobacco).
According to the latest agreement, complementary schemes’ pensions (AGIRC and ARRCO) are price-
indexed from 2019 to 2022, then indexed according to the evolution of the average wage affected by a
sustainability coefficient of - 1.16 % from 2023 onwards 2020.
Non earnings-related minimum pensions (ASPA and AAH) are also price-indexed.
Taxation
Pensions are subject to the General Social Contribution (CSG) at a rate of 8.3%, the Social Debt
Repayment Contribution (CRDS) at a rate of 0.5% and to two different health contributions: a specific
contribution for all pensioners (Additional Solidarity Contribution for Autonomy - Casa) at a rate of
0.3% and a 1% health insurance contribution withholding which is applied to both compulsory and non-
compulsory supplementary pension schemes.
Pensioners can benefit from a reduced rate of 3.8% of CSG if their income is below a certain ceiling
(€14,780 for a single tax unit in 2020 – calculated on the reference taxable income for 2018) or an
intermediate rate of 6.6% if their reference taxable income is between €14,781 and €22,940 in 2018 for
a single tax unit. They can also benefit from an exemption of CSG and CRDS if their income is below
a certain ceiling (€11,306 for a single tax unit in 2020 – calculated on the reference taxable income for
2018). Also, the Additional Solidarity Contribution for Autonomy of 0.3% is only levied for pensioners
subject to a CSG rate of 6.6% or 8.3%.
Pensions are also subject to income taxation (after a 10% rebate on the tax base). All in all, the average
tax rate in 2019 was 13.3%: 6.4% for income taxation and 6.9% for other taxes (CSG-CRDS-Casa).
There is no taxation (CSG, CRDS, Casa,income taxation) on ATMP pensions nor non earnings-related
minimum pensions (AAH, ASPA).
1.2. Recent reforms of the pension system included in the projections
Up to the last projection exercice, the French pension schemes have known five main reforms: the 1993
reform in the private sector, the 2003, 2008, 2010 and 2014 reforms that affect both private and public
sectors. Regarding the Ageing Report 2021 projections, the 2015 reform affects the AGIRC-ARRCO
schemes, thus the private sector. In addition, the reforms from 2017 to 2019 had a less significant impact
on the French pension system.
The 1993 reform introduced mainly four changes that reduced the pension level:
- The reference wage is now calculated on the basis of the 25 best wages instead of the 10 best
ones;
- Past wages factored into the calculation of the reference wage are price-indexed (and not wage-
indexed anymore);
- Pensions have become price-indexed;
- The reference contribution period has been raised from 37.5 to 40 years for private sector
employees.
12
The 2003 reform:
- It planned to semi-automatically increase the contribution period necessary to draw a full
pension in line with life expectancy gains. The aim was to keep the ratio between contribution
period and average length in retirement constant at its value of 2003 (1.79)
7
. In application of
that principle, the reference contribution period has increased from 40 years for generation 1948
to 41.5 years for generation 1957. This mechanism has been replaced by the 2014 reform.
- It created the possibility for people with long careers to retire early and scheduled an increase
of the minimum earnings-related pension. The early retirement arrangement for long careers
concerns people who started to work before the age of 16 or 20 and who have contributed longer
than the reference contribution period. They are entitled to withdraw their pension up to 4 years
before the legal retirement age (56 years old). With the legal obligation to study until the age of
16, fewer and fewer people will be eligible to this arrangement.
- A bonus system was introduced (in all schemes) for people postponing their entry into
retirement while they have reached the minimum retirement age and they meet the reference
contribution period condition. The penalty for early-retirement was gradually decreased from
10% to 5% of pension benefits for private sector workers and was introduced for the public
scheme. The reform also introduced the possibility of cumulating a pension and a wage and
fostered the development of occupational and voluntary private savings through fiscal
incentives.
- A gradual convergence of the public sector schemes toward the private sector one was
implemented through three channels: firstly, an increase of the number of contribution years
required for entitlement to a full pension (from 37.5 to 40 years); secondly, the creation of a
penalty for early retirement and a premium for postponed retirement converging gradually to
the value of the parameters in the CNAV; finally, the creation of a complementary scheme
(RAFP).
The 2008 “rendez-vous”:
- The bonus for extra years worked after having reached the required contribution period for a
full pension was raised to 1.25% per additional quarter;
- The possibility of drawing concurrently a pension and a wage was fully liberalized for people
entitled to a full pension;
- Employers were encouraged to reach quantitative targets for senior workers’ employment and
discouraged to use retirement as a substitute for layoff.
- The conditions for perceiving the Minimum Contributif (also called Mico), a contributory
minimum pension created for people entitled to a full pension, were strengthened. This
minimum pension is now means-tested in order to target people with low levels of pension more
effectively.
7
Average length in retirement is defined as the life expectancy at age 60 published five years before by the national
statistical agency (Insee). Until 2014, the COR pronounced every year (every 4 years before) a recommendation
concerning the reference contribution period that will apply to the concerned generation: everyone is therefore
informed at age 55 of the actual reference contribution length that will apply to them.
13
The 2010 reform introduced several new measures aimed at both curbing expenditures and raising
revenues:
- It introduced a progressive rise of age boundaries. The earliest retirement age was gradually
increased, for all pension schemes, from 60 to 62. Simultaneously, the full pension age went up
from 65 to 67. Every generation from generation 1951 to generation 1955 have seen these age
limits rise by 4 or 5 months
8
. For example, people born in 1956 can claim their pension at age
62 in 2018 and a full pension at 67 in 2023. The early retirement age for long careers has also
been increased by 2 years. The 2010 reform, so as the 2008 “rendez-vous” increased the
minimum contribution period required for a full pension before the age of 67.
- Exceptions related to fragile workers have been introduced. Some categories of workers are still
being granted a full pension at 65 (disabled, mother of 3 children), and people suffering from a
professional disease or an accident that resulted in a permanent incapacity of at least 20%
9
can
still retire at 60 with a full pension. The retirement for long careers is extended to people who
started to work before 18; they can retire at age 60, if they meet certain conditions.
- The convergence of pension rules between public and private sectors was strengthened by the
decision to remove the possibility of early retirement for parents with 3 children and a 15 year-
career in the public sector and the "Cessation Progressive d'Activité" programme in the public
sector as well. Rules to compute minimum earnings-related pensions and the contribution rate
of civil servants
10
will also converge towards the private sector rules.
The 2014 reform introduced short-term measures (increase of social contributions of both employ-
ees and firms by 0.3 point between 2013 and 2017, removal of the 10 % tax exemption on the pen-
sion bonus for pensioners with 3 (or more) children, postponement of the pension indexation) but
also several long-term measures:
- It introduced a progressive rise of the reference contribution period for a full pension before the
age of 67 to 43 years (reached in 2035). This rule replaces the mechanism introduced by the
2010 reform and affect all pension schemes (basic private sector schemes, the public sector
scheme, special schemes and 2nd pillar schemes);
- In order to strengthen the governance, a steering committee has been established and has been
entrusted with the task to publish a yearly report on the French pension system, including long-
term projections. It will make recommendations if there are significant discrepancies with the
baseline scenario.
The 30 October 2015 agreement on complementary pension schemes AGIRC and ARRCO intro-
duced a series of measures related to: (i) the amount of pension benefits paid to retirees, (ii) retire-
ment age, with incentives to postpone retirement, (iii) governance, with the merger of the executive
and non-executive schemes, and (iv) social contributions.
- The measures concerning the amount of pension benefits are being implemented from 2016 to
2018; part of the adjustment affects current pension recipients by restricting nominal increases
8
Initially, a 4 month increase by generation was planned between the generations 1951 and 1956 but the 2012
social security budget law planned an acceleration of this increase.
9
10% under specific disability conditions.
10
The contribution rate for civil servants will increase from 7.85% to 11.10% by 2020. It amounted to 10.29% in
2017.
14
in existing pensions, and part will affect future pensioners by making the pension system less
generous in the long run.
- Incentives to remain in employment (“solidarity coefficients” and “increase coefficients”)
should raise the effective retirement age and maintain around 100,000 additional persons in the
labour force in 2025, thus raising the amount of contributions.
- Merging the AGIRC (executive) and ARRCO (non-executive) schemes in 2019 will simplify
the pension system and reduce administrative costs.
- In the new unified scheme, the contribution assessment base will be broadened and certain
contribution rates will be increased.
In July 2017, the LURA reform (LURA stands for Liquidation Unique de retraite de base des Ré-
gimes Alignés) entered into force. Before the reform, private sector workers who had contributed to
several basic schemes over their career (CNAV, MSA salaries or SSI) used to receive as many
pensions as schemes they had contributed to and each pension was calculated separately. Since July
2017, individuals who are in such a situation
11
receive only one pension calculated according to one
single benefit formula.
- The reference wage is the average of the 25 best annual wages (valorized in line with inflation)
across the entire career.
- An individual can only validate 4 quarters per year: individuals who have contributed to two
schemes simultaneaously will get a lower pension than what they would have received before
the LURA reform.
The LURA arrangement was part of the 2014 reform but the executive order related to its
implementation was published only in May 2017.
The following reforms are new to Ageing report 2021 projections :
In May 2019, the evolution of the AGIRC-ARRCO purchase and service value of the point, decided
aims to maintain the financial balance of the scheme over a multi-year horizon.Two periods are
distinguished: from 2019 to 2022 included, the service value of the point is indexed to the projected
inflation excluding tobacco prices and without the possibility of a negative indexation; the purchase
value is indexed to the yearly average gross wage. From 2023 to 2030, the service value of the point
is indexed to the yearly average gross wage minus a sustainability coefficient equal to -1.16%, and
the purchase value to the average gross wage.
In addition, from 2018 to 2020, increases in non earnings-related minimum pensions have been
introduced : the solidarity allowance for the elderly (ASPA) has increased from €803 in early 2018
to €903 by 2020 ; the adult disability allowance (AAH), standing at €819 in early 2018, amounts
€903 by 2020.
During his presidential campaign, the President announced that he will reform the pension system in
order to unify the 42 existing schemes into a single scheme, making the French pension system more
clear and fair. This reform was under discussion in parliament before being stopped by the Covid-19
health crisis.
11
The reform does not affect individuals who had already retired before July 2017
15
1.3. Description of the actual ‘constant policy’ assumptions used in the
projection
The projections are built upon a “constant policy” principle and based on the legislation and rules as of
September 2017. As described above, the rates of return of the AGIRC-ARRCO schemes are assumed
to remain constant after the last measures of the 2015 agreement are implemented, in 2018.
In order to prevent the minimum pension expenditures (ASPA and AAH) to decline too much in the
projection exercice relatively to the poverty threshold, the ASPA and AAH ceilings are indexed on
average wages when they account for less than 50 % of the poverty threshold
12
(60% of the median
wage). The impact of this methodological change on the projection results is shown in Table 19 in
section 3.6.
The two earnings-related disability pensions (ATMP pension and PI) are wage-indexed in the
projections, as their amount mainly depend on past wages taken into account for the pensions
calculation
13
.
12
In the projection exercice, the poverty threshold is deemed to evolve practically in the same way as average wage.
13
This assumption leads to overestimate the disability pensions in the projections, since pensions that have already been
provided in a previous year are revalued with the Consumer Price Index (excluding tobacco).
16
2. Overview of the demographic and labour forces projections
2.1. Demographic development
Table 2 – Main demographic variables
2019 2030 2040 2050
2060
2070
peak
va-
lue
peak
year
chang
e
2019-
2070
Population (thousand) 67 105 68 814 69
834
70
001
69
662
69 426 7003
4,1
2047
2320,7
Population growth rate 0,2 0,2 0,1 0,0 -0,1 0,0 0,3 2020
-0,2
Old-age dependency ratio (pop 65+ / pop 20-64)
36,5 44,9 51,7 54,8 55,9 56,9 56,9 2070
20,4
Old-age dependency ratio (pop 75+ / pop 20-74)
14,2 19,4 23,8 26,9 28,0 28,7 28,8 2067
14,5
Ageing of the aged (pop 80+ / pop 65+) 30,3 32,1 36,8 40,5 42,2 43,9 43,9 2070
13,5
Men - Life expectancy at birth 80,1 81,6 83,0 84,3 85,6 86,7 86,7 2070
6,6
Women - Life expectancy at birth 86,3 87,4 88,6 89,6 90,6 91,4 91,4 2069
5,1
Men - Life expectancy at 65 20,0 20,9 21,8 22,6 23,5 24,2 24,2 2069
4,2
Women - Life expectancy at 65 24,1 24,9 25,7 26,5 27,2 27,9 27,9 2070
3,8
Men - Survivor rate at 65+ 85,9 88,2 90,0 91,5 92,7 93,8 93,8 2070
7,9
Women - Survivor rate at 65+ 92,9 94,0 94,9 95,6 96,2 96,7 96,7 2070
3,8
Men - Survivor rate at 80+ 62,1 67,1 71,1 74,8 78,0 80,9 80,9 2070
18,7
Women - Survivor rate at 80+ 79,3 82,3 84,6 86,7 88,5 90,0 90,0 2070
10,8
Net migration (thousand) 38,1 68,3 73,9 75,2 74,6 80,2 80,4 2024
42,1
Net migration over population change 0,3 0,5 1,1 -4,5 -1,9 -9,4 42,3 2047
-9,7
Source: Commission services based on Eurostat EUROPOP2018 data
Explanatory note: *This column represents a peak year, i.e. the year in which the particular variable reaches its maximum
over the projection period 2016 to 2070.
Table 2 provides an overview of the demographic development until 2070. The total size of the
population will increase until 2070 up to 69 million people, increasing until 2050 and slightly decreasing
thereafter. This global increase of the total population comes mainly from the increase in life
expectancies, mitigated by lower fertility rates in the second half of the period.
The age composition will change towards older people: the “old-age dependency” ratio which is the
share of older people (aged 65 and above) relative to the working age population (aged 20 to 64) will
increase from 36.5% in 2019 to 56.9% in 2070. Most of the increase in old-age dependency ratio will
occur before 2050: after this date, the ratio will sligthly increase because the increase in the number of
65+ people and the decrease in the numer of 20 to 64 people will be much lower. The “ageing of the
aged” ratio, which is defined by people older than 80 years old as a share of people aged 65 or above,
will first decrease until 2025, then increase the remainder of the projection period. Among the 65 years
old and older group, the age composition will thus change towards a higher share of the elderly (over
80).
17
Figure 1 – Age pyramid, comparison between 2019 and 2070
Source: Commission services based on Eurostat EUROPOP2018 data
The main differences between the age composition of the population in 2016 and 2070 are the following
ones:
- The share of people aged between 25 and 49 will be significantly lower in 2070 than in
2016.
- On the contrary, the share of people aged 69 and above will be higher in 2070 than in 2016.
Due to the dynamic fertility, the share of young people will still be high in 2070. As a whole, the age
pyramid would be flatter in 2070 than in 2016.
The comparisons between age pyramids in 2016 and 2070 are quite similar between men and women,
except that the share of the elderly will be even higher for women than for men in 2070.
2.2. Labour force
Pension reforms that shift retirement age (both early and statutory) or rise contribution period
requirements as well as active labour market policies aim to prolong working life.
43210
FR - Population by age groups and sex as a
share of total population
Males
01234
0-4
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-79
80-84
85-89
90+
Females
Age
groups
2070
2019
18
Table 3 – Participation rate, employment rate and share of workers for the age groups
20-64, 20-74, 55-64 and 65-74
2019 2030 2040 2050 2060 2070
peak
value
peak year
change
2019-
2070
Labour force participation
rate 20-64
78,0 78,8 79,7 80,0 80,1 80,0 80,2 2057 1,9
Employment rate of workers
aged 20-64
71,6 72,4 73,8 74,6 74,7 74,5 74,8 2057 2,9
Share of workers aged 20-
64 in the labour force 20-64
91,8 91,9 92,6 93,2 93,2 93,2 93,2 2057 1,5
Labour force participation
rate 20-74
66,2 66,5 67,2 68,2 68,3 68,2 68,5 2066 2,0
Employment rate of workers
aged 20-74
60,8 61,2 62,3 63,7 63,8 63,7 64,0 2066 2,9
Share of workers aged 20-
74 in the labour force 20-74
91,8 92,0 92,7 93,4 93,4 93,4 93,4 2055 1,6
Labour force participation
rate 55-64
56,9 62,9 65,8 66,4 67,3 67,0 67,3 2058 10,1
Employment rate of workers
aged 55-64
53,0 58,7 61,8 62,7 63,5 63,3 63,6 2058 10,2
Share of workers aged 55-
64 in the labour force 55-64
93,2 93,4 93,9 94,3 94,4 94,4 94,4 2065 1,2
Labour force participation
rate 65-74
5,5 9,1 11,8 14,4 14,0 14,6 14,7 2067 9,0
Employment rate of workers
aged 65-74
5,4 8,8 11,5 14,1 13,6 14,2 14,3 2067 8,8
Share of workers aged 65-
74 in the labour force 65-74
97,2 96,8 97,1 97,4 97,4 97,4 97,4 2059 0,2
Median age of the labour
force
41,0 41,0 41,0 41,0 42,0 41,0 42,0 2056 0,0
Source: Commission services
Explanatory note: *This column represents a peak year, i.e. the year in which the particular variable reaches its maximum
over the projection period 2019 to 2070.
The effects of these reforms in France are reflected in the increase of participation rate and employment
rate of the elderly (see Table 3). In line with the rise observed during the past 10 years, participation and
employment rates of the 55 to 64 years old will globally keep increasing until 2060: respectively from
56.9 % in 2019 to 67.3 % in 2060 for the participation rate, and 53.0 % to 63.6 % for the employment
rate.
Table 4a – Labour market exit age, effective retirement age and expected duration of
life spent in retirement – MEN
2020 2030 2040 2050 2060 2070
peak
value
peak
year
change
2020-2070
Average effective retirement age (admin-
istrative data)*
62,0
Average labour market exit age (CSM)** 62,3 63,6 64,5 64,7 64,7 64,7 64,7 2042 2,4
Contributory period 38,3 33,8 34,9 34,8 34,2 34,6 38,3 2020 -3,7
Duration of retirement*** 22,2 21,6 22,6 22,6 23,5 24,2 24,2 2069 2,0
Duration of retirement/contributory period
0,6 0,6 0,6 0,6 0,7 0,7 0,7 2069 0,1
Percentage of adult life spent in retire-
ment****
33,4 32,1 32,7 32,6 33,5 34,1 34,1 2069 0,8
Early/late exit***** 7,8 3,4 1,9 1,6 1,5 1,4 7,8 2020 -6,4
Source: Commission services; Insee, DESTINIE model, calculations: DG Trésor
19
Table 4b – Labour market exit age, effective retirement age and expected duration of
life spent in retirement – WOMEN
2020 2030 2040 2050 2060 2070
peak
value
peak
year
change
2020-2070
Average effective retirement age (ad-
ministrative data)*
62,6
Average labour market exit age (CSM)**
62,2 63,3 64,1 64,3 64,3 64,3 64,3 2041 2,1
Contributory period 32,4 28,8 31,0 30,7 31,6 31,1 32,4 2020 -1,2
Duration of retirement*** 26,7 26,7 26,6 27,4 28,1 28,8 28,8 2069 2,1
Duration of retirement/contributory pe-
riod
0,8 0,9 0,9 0,9 0,9 0,9 0,9 2025 0,1
Percentage of adult life spent in retire-
ment****
37,6 37,1 36,6 37,2 37,8 38,3 38,3 2069 0,7
Early/late exit***** 10,9 3,7 2,1 1,7 1,7 1,5 10,9 2020 -9,4
Source: Commission services; Insee, DESTINIE model, calculations: DG Trésor
Explanatory note: * The effective retirement age shows the age at which people on average start receiving a pension
benefit. It is calculated on the basis of the administrative data for 2019 (see Annex Tables A4a and A4b); ** The labour market
exit age as calculated based on Labour Force Survey data for the base year and estimated by the Cohort Simulation Model
thereafter; *** ‘Duration of retirement’ is calculated as the difference between the life expectancy at the average labour market
exit age and that exit age itself; **** The ‘percentage of adult life spent in retirement’ is calculated as the ratio between the
duration of retirement and the life expectancy minus 18 years; ***** Early/late exit is the ratio between those who retire and
are below the statutory retirement age and those who retire at the statutory retirement age or above.
The considerable diversity between the different schemes in terms of careers, particularly the age of
entry or demographics such as the share of women with the subject of the contribution quarters allocated
for maternity and child-rearing represent a limitation of the indicator presented in Table 4 for France.
Finally, differences in retirement age between schemes represent another limitation.
20
3. Pension projection results
3.1. Extent of the coverage of the pension schemes in the projections
The French projections cover all public pensions. Both basic and mandatory complementary schemes
have been taken into account. Given their low weight in the French pension system, occupational
pensions (with contractual agreements between employers and employees) are not covered in the
projections. Private mandatory pensions do not exist in France.
The projections cover old-age and early pensions as well as survivors’ pensions and the minimum old-
age allowance, called “ASPA” (formerly “minimum vieillesse”). They also cover disability pensions
paid before and after the minimum retirement age (including “pension d’invaliditié” (PI), allowance for
disabled adults (AAH), and “ATMP” pension for adults with a disability due to work and reducing their
capacity to work), even though they are part of health expenditures in the French accounting system.
People eligible to both ATMP pension and PI are identified from the administrative dataset of the Health
insurance schemes. Using aggregated data from the Ministry of Solidarity and Health, we estimate that
15 % of people eligible for AAH are also eligible for an ATMP pension or PI. Data on disability include
expenses and number of French people living abroad.
Table 5 - Eurostat (ESSPROS) vs. Ageing Working Group definition of pension
expenditure (% GDP)
2009 2010 2011 2012 2013 2014
2015
2016
2017
2018
2019
Change
2009-2019
1 Eurostat total pension
expenditure
14,3 14,4 14,5 14,8 15,1 15,1
15,1 15,1 14,9
2 Eurostat public pension
expenditure
3 Public pension expendi-
ture (AWG)
14,6 14,5
14,3 14,2 14,0 14,9
4 Difference (2) - (3)
5 Expenditure categories
not considered in the
AWG definition, please
specify:
5.1Early retirement
benefits
Source: Eurostat ESSPROS data (July 2017) and Social Protection Accounts, 2014 and 2018
Early retirement benefit values before 2011 are interpolated.
In this exercise like in the 2015 and 2018 ones, we strictly limit expenditures to pensions and take into
account disability pensions paid before and after the legal retirement age. Similarly to the 2018 exercise,
we include the allowance for adults with disability (AAH, 0.4% of GDP) in disability pensions, which
are not included by Eurostat and was previously in long-term care projections. This allowance was
shifted from long-term care to disability pensions due to changes in social protection accounts’
classifications.
21
Only a global projection of pension expenditures is provided in the projections, aggregating all
mandatory pension schemes for public, private and self-employed workers. The following table lists the
main pension schemes along with the amount of pensions distributed in 2018. No particular assumption
is made about the evolution of the respective shares of the different schemes.
Billion € 2018 % of GDP
CNAV
128,3 5,5 %
CNAVPL*
1,6 0,1 %
MSA employees
8,0 0,3 %
AGIRC-ARRCO
79,6 3,4 %
FPE
55,4 2,3 %
CNRACL
20,4 0,9 %
Special schemes
14
11,6 0,5 %
MSA farmers*
12,7 0,5 %
Source: Social Protection Accounts, Drees, 2020
*basic scheme only
3.2. Overview of projection results
Gross public pension spending is predicted to decrease from 14.8 % of GDP in 2019 to 12.6 % in 2070,
and peak in 2032 at 15.6 % of GDP, which represents an overall decrease of 2.2 GDP points over the
whole 2019-2070 period. Net public pension
15
spending is also predicted to decrease from 12.8 % of
GDP in 2019 to 10.9 in 2070, and peak in 2032 at 13.5 % of GDP which represents an overall decrease
of 1.9 GDP points over the whole 2019-2070 period (Table 6).
Pension contributions refer only to contributions that are collected on labour income, and not taxes on
pensions since they are not directly attributed to the financing of pension system. It must also be
highlighted that those contributions represent only a part of the global resources available: for old age
pensions, it represents around 80% of the global resources available in 2019; the remaining 20% is
collected through earmarked taxes, the FSV financial fund and taxes based on all the other types of
revenue (capital, replacement revenue…). Public pension contibutions are also predicted to decrease of
0.3 GDP points over the next five years but to remain stable around 11.5% of GDP until 2070.
14
SNCF, CRPCEN, CAVIMAC, ENIM, CANSSM, CNBF
15
Computed applying an average tax rate to the total gross public pensions.
22
Table 6 - Projected gross and net pension spending and contributions (% GDP)
Expenditure
2019 2030 2040 2050 2060 2070
Peak
value
Peak
year**
Change
2019-
2070
Gross public
pension
expenditures
14,8% 15,6% 15,2% 14,3% 13,4% 12,6%
16,2% 2020
-
2,2
Pivate
occupational
pensions
: : : : : :
:
:
Private
individual
mendatory
pensions
: : : : : :
:
:
Private
individual non
mendatory
pensions
: : : : : :
:
:
Gross total
pension
expenditures
14,8% 15,6% 15,2% 14,3% 13,4% 12,6%
16,2% 2020
-
2,2
Net public
pension
expenditures*
12,8% 13,5% 13,2% 12,4% 11,6% 10,9%
14,1% 2020
-
1,9
Net total
pensoin
expenditures*
12,8% 13,5% 13,2% 12,4% 11,6% 10,9%
14,1% 2020
-
1,9
Contributions
2019 2030 2040 2050 2060 2070
Peak
value
Peak
year**
Change
2019-
2070
Public pension
contributions
11,8% 11,5% 11,5% 11,5% 11,6% 11,6%
12,2%
2020
-
0,2
Total pension
contributions
11,8% 11,5% 11,5% 11,5% 11,6% 11,6%
12,2%
2020
-
0,2
Source: Commission services; Insee, DESTINIE model
Explanatory note: *net pension expenditure excludes taxes on pensions and compulsory social security contributions paid by
beneficiaries.**This column represents a peak year, i.e. the year in which the particular variable reaches its maximum over
the projection period 2019 to 2070.)
Concerning the 2019-2070 evolution of the ratio between projected gross pension expenditures and
GDP, six periods can be identified (Graph 2):
• Phase 1: Following the Covid-19 crisis, GDP was revised downwards which mechanically
increased the weight of pension expenditure to GDP. Up to 2021, pension expenditures as a
share of GDP will mainly reflect macroeconomic forecasts too.
• Phase 2: From 2021 to 2024, pension expenditures as a share of GDP will slightly decrease from
15.4% to 15.3% despite the ageing of the population due to the progressive increase of the
retirement age (past reforms).
• Phase 3: From 2025 to 2032, the ratio between pension expenditures and GDP will increase up
to 15.6% and then stagnate around this level until 2032. During this period, the number of
pensioners and the amount of new pensions is expected to continue raising at a sustained pace.
The ratio of pension expenditures to GDP is expected to decline continuously over the period 2033-2070,
to reach 12.5 % of GDP in 2070. This decline can be further broken down into three sub-periods:
23
• Phase 4: From 2033 to 2042, growth in pension expenditures will start showing signs of
deceleration which reflects the effect of the 2014 reform which consists in a progressive increase
of the minimum contribution period for a full rate pension.
• Phase 5: From 2043 to 2063, pension expenditures as a share of GDP are expected to decline
due to the slowdown of the ageing process which will prevent the number of people aged 65
and above and hence the number of retirees from growing. Strong nominal GDP growth relative
to growth in pension expenditures will also contribute to the sustained decline in the public
pension expenditure-to-GDP ratio.
• Phase 6: From 2063 to 2070, pension expenditures as a share of GDP will continue to decline
but at a slower pace as the growth rate of people aged 65 and above will accelerate and nominal
GDP growth will slightly decelerate.
Figure 2 - Projected pension expenditures (% of GDP)
Source: Insee, DESTINIE model, calculations: DG Trésor
11.0%
12.0%
13.0%
14.0%
15.0%
16.0%
17.0%
2019 2024 2029 2034 2039 2044 2049 2054 2059 2064 2069
Phase
2
Phase 3
Phase 4
Phase 5
Phase 6
Phase
1
24
Table 7 - Projected gross public pension spending by scheme (% GDP)
Pension scheme
2019 2030 2040 2050 2060 2070 Peak value Peak year *
Change
2019-2070
Total public pensions
14,8% 15,6% 15,2% 14,3% 13,4% 12,6%
16,2%
2020
-2,2
Old age and early pensions:
12,1% 13,1% 12,8% 12,1% 11,4% 10,7%
13,4%
2020
-1,4
Flat component
: : : : : :
:
:
:
Earnings related
12,0% 12,8% 12,5% 11,8% 11,1% 10,5%
13,1%
2020
-1,5
Minimum pensions (non-contribu-
tory) i.e. minimum income guarantee
for people above 65
0,2% 0,2% 0,3% 0,3% 0,2% 0,2%
0,3%
2040
0,1
Disability pensions
1,10% 1,08% 1,03% 0,97 % 0,95% 0,95%
1,2%
2020
-0,1
Survivor pensions
1,53% 1,46% 1,36% 1,18% 1,06% 0,94%
1,67%
2020
-0,6
Other pensions : : : : : : : : :
Source: Commission services; Insee, DESTINIE model
Note : Old age and early pensions include special sheme pensions
As old-age earnings-related pension expenditures represent the major share of total pension
expenditures, they follow the same evolution as described in 3.1.2. section : old-age earnings-related
pension spending (Table 7), as a share of GDP, would go down from 12.1 % of GDP in 2019 to 10.7 %
of GDP in 2070. By shifting the legal and statutory retirement ages, and increasing the minimum
contribution period, the 2014 reform contributed to reduce the weight of total pension expenditures in
GDP. Moreover, more fragmented careers and later entry into the labour market imply a lower average
amount of pensions as well as a lower average replacement rate at retirement.
The ratio between non earnings-related old-age pensions (“ASPA” or “minimum vieillesse”) and GDP
will first increase until 2037 and then slightly decrease to 0.2 % by 2070 (the impact of the change in
the indexation mechanism from 2063 onwards (wage instead of CPI indexation) is relatively small (cf.
Table 19).
Survivors’ pensions, as a share of GDP, are expected to decline from 1.1 % in 2019 to 1.0 % of GDP in
2070. The overwhelming majority of survivor pensions’ beneficiaries are women: the reduction of the
gap between life expectancies of men and women, the relative increase of women employment rates,
and the decrease of the number of weddings induce that women will have a lower and time-limited
amount of survivors’ pensions over the projection period.
Non-earnings-related minimum disability pension expenditures (AAH), as a share of GDP, are projected
to decline slightly until 2063 in the baseline scenario as AAH is price-indexed at first
16
.
16
The government has decided to increase the AAH (non-earnings-related disability pension) on an extraordinary basis in
November 2018 and November 2019 in order to bring the amount of the benefit to 900 euros by 2019 from 810,89 euros in
2017. -
25
3.3. Description of main driving forces behind the projection results
and their implications
In order to identify more clearly the driving forces behind the projection results, the pension-to-GDP
ratio is split into 4 factors:
Figure 2 – Disaggregation of public pension expenditure
Figure 3 - Contribution of the main driving forces to the projection results
Source: Insee, D
ESTINIE model, calculations: DG Trésor
=
65+
20−64
×
65+
×
ℎ 20−74
×
20−64
ℎ 20−74
[
1
]
65+
=
65+
65+
+
≤65
50−64
×
50−64
65+
[2]
20−64
ℎ 20−74
=
20−64
20−64
×
20−64
ℎ 20−64
×
ℎ 20−64
ℎ 20−74
[3]
d
ependency rati
o
coverag
e
rati
o
l
abour market effec
t
benefi
t
rati
o
coverag
e
rati
o
ol
d
-ag
e
coverag
e
rati
o
earl
y
-ag
e
c
ohort effec
t
1/employment rat
e
1/labour intensit
y
1/career shif
t
-10%
-8%
-6%
-4%
-2%
0%
2%
4%
6%
8%
2019 2024 2029 2034 2039 2044 2049 2054 2059 2064 2069
Dependancy ratio Labour intensity Coverage ratio
Benefit ratio Public pensions to GDP
26
On the basis of the Eurostat demographic assumptions, the dependency ratio (population 65+/population
20-64) notably increases up to 2055 and then remains stable until 2070 (Figure 3). Concerning the
impact, ceteris paribus, of each of the factors considered on the evolution of pension expenditures (Table
8.a), the dependency ratio pushes up pension expenditures between 2019 and 2070 (+6.5 pp).
The coverage ratio (pensioners/population 65+) regularly decreases until 2040 and then stabilizes up to
2070. This is linked to the increase in retirement ages planned by the 2010 reform, but also to the increase
in the full pension contribution period planned by the 2014 reform which, associated with an increase
in the labour market entry age, leads to higher retirement ages. The coverage ratio mainly reduces public
pension expenditures as a share of GDP until 2040 (-1.3 pp).
The benefit ratio (defined as the average pension benefit divided by the economy-wide average wage)
declines all along the period (except in 2020), to reach in 2070 a level which is 66% lower than the
current level. The reduction of the benefit ratio reflects the subdued growth pace of the average pension
compared to that of the average wage per worker. First, the increase in discontinuous careers due to high
unemployment rates will not only decrease the average of the 25 best yearly wages (used to calculate
the pension) but also impact the prorata coefficient if individuals do not have the required number of
contributed years. Second, pensions are price-indexed while the average wage per worker increases in
line with labour productivity or GDP per worker, hence more rapidly. Finally, changes to the benefit
formulae induced by the last reforms will also contribute to the decline of the benefit ratio.
The labour market indicator (population aged 20 to 64/employed population aged 20 to 74) declines
over the first 30 years of the projection horizon and remains quite stable thereafter.
Table 8 - Factors behind the change in public pension expenditure between 2019 and
2070 (percentage points of GDP) – pensioners
S
2019-30 2030-40 2040-50
2050-
60
2060-
70
2019-
70
Public pensions to GDP
0,8 -0,4 -0,9 -0,9 -0,8 -2,2
Dependency ratio effect
3,4 2,3 0,9 0,3 0,2 7,1
Coverage ratio effect*
-1,1 -0,6 -0,1 -0,1 -0,1 -2,0
Coverage ratio old-age -0,1 -0,1 0,0 0,0 0,0 -0,2
Coverage ratio early-age -1,3 -1,7 0,0 -0,5 -0,2 -3,7
Cohort effect
-3,0 -2,2 -1,1 0,0 -0,1 -6,5
Benefit ratio effect
-1,0 -1,6 -1,4 -1,1 -0,9 -5,9
Labour market effect
-0,3 -0,4 -0,2 0,0 0,0 -1,0
Employment ratio effect -0,2 -0,3 -0,2 0,0 0,0 -0,6
Labour intensity effect 0,0 0,0 0,0 0,0 0,0 0,0
Career shift effect -0,2 -0,1 -0,1 0,0 0,0 -0,4
Residual
-0,2 -0,1 0,0 0,0 0,0 -0,4
Source: Commission services; Insee, DESTINIE model, calculations: DG Trésor
* Sub components of the coverage ratio effect do not add up necessarily.
The benefit ratio (BR - calculated as the average pension compared to the economy-wide average wage)
and the replacement rate (RR - calculated as the ratio between the average pension of new pensioners
and the average wage at retirement) will both decline during the projection horizon (Table 09). This
decline comes from several factors: the rise in the full pension contribution period, the rule used to
discount past wages entering into the benefit formula in interaction with an increase in the labour market
27
entry age and more fragmented careers, the development of polypension (when one pensioner cumulates
several pensions) that can imply smaller pensions because of the specific rules applied in this situation.
RR are usually calculated individually, by comparing the new pension at retirement with the last wage
at retirement
17
, and presented through the median replacement rate of the population. They are also often
calculated for theoretical or typical careers, for instance an entire career of a private sector employee
paid at the average wage. On the contrary in this exercise, RR are averaged over all careers and all
schemes; they are therefore smaller than the replacement rates exhibited in other reports, and should be
interpreted carefully. The BR, calculated by comparing pensions and wages of different generations, is
not a replacement rate: its evolution reflects the relative differences in the standards of living of the
workers and the pensioners.
Table 9 - Benefit ratio (BR), Replacement rate at retirement (RR) and Coverage by
pension scheme (in %)
2019 2030 2040 2050 2060 2070
change
2019-2070
(pps)
Public scheme (BR) 41% 39% 35% 32% 30% 28% -13%
Coverage 61,3 58,5 57,3 56,5 55,7 55,3 -6,0
Public scheme: old-age earnings re-
lated (BR)
52% 50% 45% 41% 38% 35% -17%
Public scheme: old-age earnings re-
lated (RR)
54% 50% 40% 42% 37% 35% -20%
Coverage 38,7 37,9 37,1 36,9 36,6 36,6 -2,1
Private occupational scheme (BR) : : : : : : :
Private occupational scheme (RR) : : : : : : :
Coverage : : : : : : :
Private individual schemes (BR) : : : : : : :
Private individual schemes (RR) : : : : : : :
Coverage : : : : : : :
Total benefit ratio : : : : : : :
Total replacement rate 0 0 0 0 0 0 0%
Source: Commission services ; Insee, DESTINIE model, calculations: DG Trésor
Explanatory note: Coverage of each pension scheme is calculated as a ratio of the number of pensioners within the scheme
and the total number of pensioners in the country.
Like in the 2018 exercise, there are two reasons why the number of pensioners is higher than the number
of people aged 65 and older (Table 10): on the one hand, some of the pensioners are younger than 65.
On the other hand, pensioners living abroad are included, while the demographic projections are limited
to the French territory
18
. In addition, pensioners living abroad are fewer in the 2021 projections than it
was in 2018.
The number of pensioners increases by 25% between 2019 and 2070, versus 0,3% only for the employed
population (Table 10). The growth of the number of pensioners is mostly concentrated before 2040, in
line with the demographic projections. This leads to an increase in the retired-to-employed population
17
Nonetheless, the definition of the replacement rate varies over the sources, and especially the definition of the
reference wage. There are many different publications which compare the new pension with the last full-time
wage, the average last 5 yearly wages, the wage at 50 years old, etc.
18
As a matter of fact, the sample of the population used to feed the Destinie model includes people living in France
only, but pensioners living abroad are included ex-post.
28
ratio between 2019 and 2070, with a stronger increase before 2040. The old-age dependency ratio
follows the same trend. The system efficiency ratio is expected to decrease over the projection period,
mainly due to the evolution of the old-age dependency ratio.
Table 10 – System Dependency Ratio and Old-age Dependency Ratio
2019 2030 2040 2050 2060 2070
change
2019-2070
Number of pensioners
(thousand) (I)
24079,9 27326,2 29694,4 30748,1 30925,9 30935,6
6855,7
Employment (thousand) (II)
27566,1 27869,4 28010,2 28016,8 27826,1 27641,2
75,1
Pension system depend-
ency ratio (SDR) (I)/(II)
87,4 98,1 106,0 109,7 111,1 111,9 24,6
Number of people aged
65+ (thousand) (III)
13611,4 16555,4 18680,9 19443,9 19724,0 19942,3
6331,0
Working age population 20-
64 (thousand) (IV)
37326,7 36906,3 36101,7 35510,6 35255,3 35050,7
-2275,9
Old-age dependency ratio
(OADR) (III)/(IV)
36,5 44,9 51,7 54,8 55,9 56,9 20,4
System efficiency
(SDR/OADR)
2,4 2,2 2,0 2,0 2,0 2,0 -0,4
Source: Commission services; Insee, DESTINIE model, calculations: DG Trésor
With regard to the age decomposition (Tables 11.a to 12.b), one should note that before the age of 60,
the ratio of pensioners to inactive population and total population is below 100% because there are few
possibilities to retire before 60. Around 90% of the pensioners younger than 54, and 70% of the
pensioners between 55 and 59 years old are disability pensioners. Conversely, this ratio is generally
above 100% for individuals aged 60 or above partly because inactive population is estimated on the
French territory while pensioners living abroad are still included in the projection (they account for
approximately 6% of total pensioners). Moreover, the computation of the pensioners-to-inactive
population and pensioners–to-total population ratios by age groups rely on two different data sources.
The number of pensioners by age groups is calculated based on national projections (old-age pensions
and disability pensions). Inactive and total population figures stem from labour force projections
obtained through the CSM method run by the Commission. Compared to the 2018 projection exercise,
there is a better match between CSM projection of the labour force exit age and the pension age projected
by the French microsimulation model (Graph 4.c). The pension age for men is slightly lower than the
labor market exit age calculated by the Commission. This is due to the early retirement scheme for long
careers which is not factored into the CSM calculation.
Figure 4c - Average labour market exit age (CSM) vs pension age (Destinie)
29
Source: Commission services, Insee Destinie model, calculation: DG Trésor
The pensioners-to-inactive population ratios by age groups are broadly stable over the projection period,
except for the age group 60-64. The ratio for the age group 60-64 declines by 20pp between 2016 and
2070. Although we should remain very cautious in the interpretation of this ratio and its evolution given
the aforementioned limitations, the evolution of the ratio likely reflects the effect of the 2010 and 2014
reforms that have increased the retirement age and the conditions for being granted a full rate pension.
Hence, a larger share of people aged between 60 and 64 is expected to be working in the coming decades.
This is also reflected by the evolution of the share of 60 to 64 year old pensioners among the total 60 to
64 year old population which is projected to sharply decrease over the projection horizon. The same
trend holds for women (Tables 13.a and 13.b).
The coverage ratio profiles also depend on retirement behaviour assumptions. But as the French pension
system is almost actuarially neutral at the margin, the impact of this assumption on public pension
expenditures is small (cf. annex E).
30
Table 11A – Pensioners (public schemes) to inactive population ratio by age group (%)
2019 2020 2030 2040 2050 2070
Age group -54 8,4 8,5 8,3 8,2 8,1 8,1
Age group 55-59 88,7 87,3 90,5 84,3 82,0 77,3
Age group 60-64 93,8 92,1 95,0 88,8 88,6 86,3
Age group 65-69 109,3 108,2 112,2 113,3 117,9 116,1
Age group 70-74 107,7 107,7 107,3 108,3 110,1 110,2
Age group 75+ 104,3 104,5 105,9 105,9 105,7 105,7
Source: Commission services; Insee, DESTINIE model, calculations: DG Trésor
Table 11B – Pensioners (public schemes) to population ratio by age group (%)
2019 2030 2040 2050 2060 2070
Age group -54 3,8 3,7 3,6 3,6 3,7 3,6
Age group 55-59 20,3 20,6 20,1 19,2 18,9 18,1
Age group 60-64 60,6 49,3 40,1 38,3 37,5 36,7
Age group 65-69 100,8 95,0 90,9 90,2 89,6 88,4
Age group 70-74 104,6 104,8 104,6 105,1 104,6 105,1
Age group 75+ 104,3 105,9 105,9 105,7 105,8 105,7
Source: Commission services; Insee, DESTINIE model, calculations: DG Trésor
Table 12A – Female pensioners (public schemes) to inactive population ratio by age
group (%)
2019 2030 2040 2050 2060 2070
Age group -54 8,0 7,9 7,8 7,7 7,8 7,4
Age group 55-59 83,7 84,9 80,5 79,6 76,6 72,6
Age group 60-64 92,6 92,6 91,4 90,3 90,5 89,2
Age group 65-69 106,0 107,9 111,6 115,3 116,4 115,9
Age group 70-74 105,9 106,8 108,1 110,0 108,4 110,6
Age group 75+ 104,1 105,7 105,6 105,4 105,6 105,2
Source: Commission services; Insee, D
ESTINIE model, calculations: DG Trésor
Table 12B – Female pensioners (public schemes) to total population ratio by age group
(%)
2019 2030 2040 2050 2060 2070
Age group -54 3,8 3,6 3,5 3,5 3,5 3,4
Age group 55-59 21,8 22,1 21,5 20,1 19,3 18,3
Age group 60-64 60,7 50,7 43,9 41,8 40,9 40,3
Age group 65-69 99,2 93,2 90,6 89,4 89,8 89,2
Age group 70-74 103,8 104,5 104,6 105,2 103,8 105,6
Age group 75+ 104,1 105,7 105,6 105,4 105,6 105,2
Source: Commission services; Insee, DESTINIE model, calculations: DG Trésor
The flow of new pension expenditures (public old-age earnings-related pensions for new pensioners)
can be broken down as the product of three terms: the average amount of new pensions, the number of
31
new pensioners
19
and the average number of months paid in the first year. The average amount of new
pensions can also be analyzed as the product of three terms (table 14.a)
20
:
1. the average contribution period of new pensioners ;
2. the value of pensionable earnings of new pensioners computed as the average of the present
value of the 25 best annual wages
21
;
3. the effective average accrual rate for new pensioners There is no administrative accrual rate
in the French legislation. Nevertheless, given the average amount of new pensions, the av-
erage contribution period among of new pensioners and the value of pensionable earnings
of new pensioners, it is possible to estimate ex-post “effective” accrual rates (see annex F).
There is no sustainability factor in the French pension system, therefore this factor remains constant
over the projection period.
The projected new pension expenditure is higher for men when compared to women (up to 24.6 M€ for
men vs. 20.1 M€ for women by 2070) even if women are more represented among new pensioners. This
is explained by the fact that women validate fewer quarters because of chopped careers and receive on
average lower salaries which impact negatively their average pension when compared to men.
Table 13a - Disaggregation of new public pension expenditure (old-age and early
earnings-related pensions) – Total
2019 2030 2040 2050 2060 2070
Projected new pension expenditures (million EUR)* 4 014 7 481 11 601 16 972 20 621 31 831
I. Number of new pensioners ('000) 621,9 804,4 772,7 803,3 698,4 737,8
II. Average contributory period 33,0 31,1 32,8 32,7 32,8 33,0
III. Average accrual rate (%) 1,1 1,2 1,1 1,1 1,0 1,0
IV. Monthly average pensionable earnings (€ ) 3 077 3 633 5 5848 7 490 10 656 15 244
V. Sustainability/Adjustment factor 1 1 1 1 1 1
VI. Average number of months paid the first year 5,9 6,6 7,6 8,0 8,1 8,2
(Monthly average pensionable earnings) /
(Monthly economy-wide average wage)
101,7% 93,7% 104,0% 98,3% 98,4% 99,1%
Source: Insee, DESTINIE model, calculations: DG Trésor
Explanatory note: Monthly average pensionable earnings are calculated as the average of the present value
22
of the 25 highest
annual wages of each individual.
*New pension expenditure equals the product of I, II, III, IV, V & VI
19
As noted previously, it is common for a French pensioner to receive several pensions due to the design of the pension system.
Therefore, we do not use the number of new pensions (as recommended by the Commission) but the number of new pensioners
for all the calculations.
20
In table 14 (a,b and c), point system schemes pensions are decomposed as if they were computed using the defined benefit
schemes formula and added to the decomposition of DB pensions. For the breakdown of new public pension expenditures by
type of scheme, cf. annex F.
21
In practice, the reference wage defined in the legislation depends on the sector considered: the 25 best years wage average is
used in the general scheme, whereas the whole career wages are used to acquire points in the complementary pension scheme,
and in the public service scheme, the reference wage is the last 6-month wage (excluding bonuses). By convention for the new
pension decomposition (but not in the pension calculation), the 25 best years wage average has been retained for all pension
schemes.
22
Past wages are are valorized in line with CPI.
32
Table 13b - Disaggregation of new public pension expenditure (old-age and early
earnings-related pensions) – MALE
2019 2030 2040 2050 2060 2070
Projected new pension expenditures (million EUR)* 2 074 3 776 5 771 9 052 10 648 18 436
I. Number of new pensioners ('000) 284,2 380,4 353,0 391,3 326,8 384,7
II. Average contributory period 37,0 33,8 34,9 34,8 34,2 34,6
III. Average accrual rate (%) 0,9 1,1 0,9 0,9 0,9 0,9
IV. Monthly average pensionable earnings (€) 3 915 4 209 6 463 8 658 12 271 18 360
V. Sustainability/Adjustment factor 1 1 1 1 1 1
VI. Average number of months paid the first year 5,9 6,6 7,9 8,3 8,6 8,6
Monthly average pensionable earnings /
Monthly economy-wide average wage
129,4% 108,6% 120,4% 113,6% 113,3% 119,4%
Source: Commission services, Insee Destinie model, calculation: DG Trésor
*New pension expenditure equals the product of II, III, IV, V, VI & VII
Table 13c - Disaggregated new public pension expenditure (old-age and
early earnings-related pensions) – FEMALE
2019 2030 2040 2050 2060 2070
Projected new pension expenditures (million EUR)* 1 940 3 705 5 829 7 920 9 973 13 395
I. Number of new pensioners ('000) 337,7 424,1 419,7 412,1 371,6 353,0
II. Average contributory period 29,5 28,8 31,0 30,7 31,6 31,1
III. Average accrual rate (%) 1,4 1,5 1,2 1,3 1,2 1,3
IV. Monthly average pensionable earnings (€) 2 347 3 105 4 841 6 363 9 233 11 846
V. Sustainability/Adjustment factor 1 1 1 1 1 1
VI. Average number of months paid the first year 5,9 6,7 7,4 7,6 7,6 7,7
Monthly average pensionable earnings /
Monthly economy-wide average wage
77,6%
80,1%
90,2%
83,5%
85,3% 77,0%
Source: Commission services, Insee Destinie model, calculation: DG Trésor
*New pension expenditure equals the product of II, III, IV, V, VI & VII
We do not consider the number of pensions but the number of pensioners. Indeed, it is possible in France
to be a poly-pensioner and to receive more than one pension from different schemes. It is then more
realistic to take into account the number of pensioners instead of number of pensions.
33
Also, as we said in the introduction of the country fiche, French pensions are build by a Defined Benefit
and a Point system. As we take into account the number of pensioners instead of number of pensions
we have to only consider the new pension expenditures related to Defined Benefits (DB) systems and
not Point systems (PS).
1. The contribution period, which equals here the number of years a person earns a labour income, is
stable over the horizon of the projection (decreasing for men and increasing for women). The de-
layed entry in the job market due to the increasing duration of studies (Graph 4.a) balances the
increase in the required contribution period. The increase in the duration of education for post-war
generations was accelerated by some policy changes, like the increase of the minimum age of man-
datory education from 14 to 16 for children born after 1953
23
. The distribution of ages of the new
pensioners does move up for both men and women between 2019 and 2070, reflecting the effects of
the recent pension reforms (Graph 4.b).
2. Pensionable earnings follow the progression of wages of individuals along their career. The increase
of the average amount of monthly pensionable earnings is thus linked to productivity gains.
3. The average accrual rate gives an insight of the ratio between the average replacement rate at retire-
ment, and the average contributory period for the entire career. Its value is higher for women than
for men mainly for three reasons:
a) There is a contributory minimum pension in the private sector as well as in the public
one. It is provided by the main pension schemes and it should not be confused with the
"Allocation de Solidarité pour les Personnes Âgées", which is a social assistance benefit
financed by the public old-age solidarity fund (FSV – Fonds de Solidarité Vieillesse). The
contributory minimum pension is attributed to people who meet the conditions for a full
pension. It is called “Minimum contributif” (or Mico) for private sector employees, “Mini-
mum garanti” (or Mingar) for public sector employees. This minimum pension benefits to
people who have earned low revenues (and/or who have worked part-time). Thus, benefi-
ciaries from this minimum pension have a relatively higher accrual rate, since they receive
a higher pension compared to what they have contributed for. A bit less than one fifth of
private sector employees are entitled to the contributory minimum. Around two thirds
of women are entitled to the contributory minimum against only 4 men out of 10, as women
have lower revenues on average. This leads to a higher average accrual rate for women.
b) Women also tend to benefit more from other redistributive elements than men (especially
maternity leave bonuses) which raises their average accrual rate compared to men.
c) By design, high-wage earners tend to have a lower accrual rate in France as only revenues
below the social security ceiling are factored into the calculation of the benefit. Thus, pen-
sions of high-wage earners are lower relative to their wage for workers with high salaries
than for low-income earners, which implies a lower accrual rate. Since wages are on average
higher for men than for women, it contributes to a lower accrual rate for men.
Finally, the number of new pensioners is not expected to increase significantly over the projection period
and is expected to remain close to the number of new pensioners per year observed over the past 10
years (cf. Table 14.a). The population is expected to age at a relatively fast pace over the first half of the
projection period (cf. evolution of the old-age dependency ratio) but the effect of the 2014 reform
23
Executive order of 6 January 1959
34
(progressive increase of the minimum contributory period required for being granted a full pension
before the age of 67) will keep the number of new pensioners contained
24
. During the second half of the
projection period (2035-2070), the population ageing process is expected to slow down, keeping the
number of new pensioners per year relatively stable.
Figure 4 - New pensioners – Breakdown by age and sex
Source: Insee, DESTINIE model, calculations: DG Trésor
24
The projections are made upon the assumption that individuals retire as soon as they are entitled to a full pension.
35
Table 13.a1 Number of new pensioners 2004-2018
in '000 Men Women All
2004
422 326 748
2005
386 330 716
2006
416 373 789
2007
427 398 825
2008
429 413 842
2009
351 388 739
2010
371 407 778
2011
313 368 681
2012
298 307 605
2013
374 384 758
2014
348 354 702
2015
327 326 653
2016
323 320 643
2017
353 354 707
2018
360 385 745
Source: La retraite et les retraités, Panoramas de la Drees, 2020
For the breakdown of new public pension expenditures by type of scheme, cf. annex F.
3.4. Financing of the pension system
In 2020, contribution rates to the general basic pension scheme stand at 10.45% of the gross wage below
the Social Security Ceiling (1 SSC = € 3 428 per month in 2020) for employers and 7.30% for workers
in the main basic scheme.
Besides contributing to the main basic scheme, executive and non-executive private sector workers
contribute to AGIRC-ARRCO at a 3.15% rate on the basis of the part of their wage below one SSC (the
contribution rate is 4.72% for the employer), and at a 8.64% rate for the part of their wage between one
and eight SSC (respectively 12.95% for their employer).
Non-executive workers also contribute to AGFF at a 0.80% rate (1.20% for their employer) on the basis
of the part of their wage below one SSC and at a 0.90% rate for the part of their wage between one and
three SSC (respectively 1.30% for their employer). Executive employees contribute to the general
scheme, to ARRCO (with respect to wage up to the ceiling), to AGFF, to another exceptional
complementary contribution CET and to AGIRC (for wage between 1 and 8 times the ceiling).
Civil servants’ contribution rate is 11.10% (employee) of their gross wage. In reality (and not in the
projections, cf. infra), the contribution rate of the State is determined and adjusted every year so as to
balance the public schemes.
36
Table 14 – Financing of the system
in 2020
NB: 2020 Social Security Ceilling
(SSC) : 41 136€ / 2020 gross mini-
mum wage: 18 473€
Public sector employees* Private sector employees** Self employed***
Basic scheme
Complementary
scheme
Basic scheme
Complementary
scheme****
Basic scheme
Complementary
scheme
Contribution base
Gross salary (traitement
indiciaire + NBI).
Excluding bonuses.
Bonuses, up to 20% of
"traitement indiciaire"
Gross annual salary including
some
types of bonuses
Gross annual salary
including bonuses
Non-salaried
work-related gross in-
come
Contribution rate 85,38% 10,00% 17,75%
7,87% up to 1 SSC,
21,59% from 1 to 8 SSC
10,10%
The contribution rate
varies depending on
type of activity
Employer
: : 10,45%
4,72% up to 1 SSC,
12,95% from 1 to 8 SSC
: :
Employee
11,10% 5,00% 7,30%
3,15% up to 1 SSC,
8,64% from 1 to 8 SSC
: :
State
74,28% 5,00% : : : :
Other revenues
Pensions reserve fund and old-age solidarity fund Pensions reserve fund and old-age solidarity fund
Pensions reserve fund and old-age solidarity
fund
Maximum contribution (annual in
2020)
:
The contribution base
cannot be higher than
20% of the gross sal-
ary
(traitement indiciaire)
There is no ceilling: for reve-
nues above the
SSC, the contribution amouts
to 2,3%
27 424 € 205 680 € :
Minimum contribution (annual in
2020)
: :
In order to validate one quar-
ter per year of
contribution (the minimum
that can be validated per year),
an employee must earn at least
150 times the hourly mini-
mum wage (1522,5€ in 2020)
over the year.
: 478 € :
*Militrary excluded, **non-executive employee earning less than the social security ceiling,
***self-employed earning less than the social security ceiling, craftsmen, tradesmen and lawyers excluded,
****AGFF included. Source DG Trésor
37
Only the contributions strictly speaking (i.e. collected on labour income) have been projected, in ac-
cordance with AWG guidelines (Table 15). As requested by the Commission, the implicit contribution
rates are kept constant in the projection interval: as a result, the share of employer and employee contri-
butions will remain stable. The State also pays a contribution as the employer of civil servants.
The number of contributors is a little bit higher than the number of employment. Indeed, in France there
is a pension called AVPF, offered by the State to people taking care of children or disabled people. This
help insures old-age earning rights to AVPF pensioners. Therefore, they are not part of the employment
but still of the contributors ( which is paid actually by the State).
Table 15 – Revenue from contribution (EUR millions), number of contributors in the
public scheme (in 1000), total employment (in 1000)
2019 2030 2040 2050 2060 2070 change
2019-2070
(pps)
Public pension contributions
(%GDP)
11,8 11,5 11,5 11,5 11,6 11,6 -0,2
Employer contributions 5,5 5,6 5,7 5,7 5,7 5,7 0,2
Employee contributions 3,8 3,8 3,9 3,9 3,9 3,9 0,1
State contribution* 2,5 2,1 2,0 2,0 2,0 2,0 -0,5
Other revenues* 0,0 0,0 0,0 0,0 0,0 0,0 0,0
Number of contributors (I)
(1000)
28322,4 28960,3 29268,7 29236,4 29125,4 28997,5
675,1
Employment (II) (1000) 27566,1 27869,4 28010,2 28016,8 27826,1 27641,2
75,1
(I) / (II) 1,0 1,0 1,0 1,0 1,0 1,0 0,0
Source: Commission services; Insee, D
ESTINIE model, calculations: DG Trésor
*only legislated contributions are reported
3.5. Sensitivity analysis
In order to assess the sensitivity of pension schemes to different economic assumptions, 12 sensitivity
tests have been carried out. Definitions of these sensitivity tests and graphs of the evolution of pension
expenditures under these scenarios are given in appendix G.
• Higher life expectancy scenario: public pension expenditures as a share of GDP are 0.6 point higher
in 2070 than in the baseline scenario (Table 17). In this scenario, pensioners live longer and earn a
pension during a longer period.
• Under the higher productivity scenario, pension expenditures are driven up as a result of higher
productivity assumption but GDP increases even more: impact on the public pension expenditure-
to-GDP ratio is negative, since in 2070 the ratio is expected to be lower by 1 point than in the baseline
scenario. On the contrary, under the lower productivity scenario, the public pension expenditure-to-
GDP ratio is 1 point higher in 2070 than in the baseline scenario.
• Higher employment rate of older workers scenario: the public pension expenditure-to-GDP ratio is
0.3 point lower than in the baseline scenario in 2070. Pension expenditures are higher in this scenario
as older workers have better carreers but the increase in pensions is more than compensated by an
higher GDP.
• Migration-related scenarios:
o Under the higher migration scenario, the public pension expenditure-to-GDP ratio is
0.3 point lower than in the baseline scenario in 2070.
38
o Under the lower migration scenario, the public pension expenditure-to-GDP ratio is 0.2
point higher than in the baseline scenario in 2070.
• Under the low fertility scenario, the smaller cohorts after 2019 lead to a lower labour force after the
mid-thirties, which decrease GDP and raise the pension expenditure-to-GDP ratio by 1.8 point in
2070.
• Covid-19 scenarios:
o Under the temporary shock scenario, the public pension expenditure-to-GDP ratio is
0.1 point higher than in the baseline scenario in the short term, before stabilizing in the
mid-term and recovering in the long term as a result of lower public pension
expenditures.
o Under the permanent shock scenario, there is a sustainably lower potential growth
which implies an increase in public pension expenditure-to-GDP ratio also in the
medium and long term. Pension expenditures and GDP are negatively affected by the
shock, but the effect on GDP dominates.
• The Linking retirement age to change in life expectancy scenario links the minimum and statutory
retirement age to increases in life expectancy after 2022, when the 2010 reform reach its full effect.
The reference contributory period is unchanged. The increase in the effective retirement age leads to
lower pension expenditures, the average pension being slightly higher but served on a shorter period.
On the contrary under the unchanged retirement age scenario public pension expenditure-to-GDP
ratio is higher than in the baseline scenario.
• Declining benefit ratio scenario: average pension to average earned income ratio is limited to decline
less than 10% over the projection horizon when compared to 2019, which leads to an increase in the
pension expenditure-to-GDP ratio by 3.4 points in 2070.
Table 17 - Public and total pension expenditures under different scenarios (pps
deviation from the baseline)
2019 2030 2040 2050 2060 2070
Change
2019-
2070
Baseline (% GDP) 14,8% 15,6% 15,2% 14,3% 13,4% 12,6% -2,2%
Higher life expectancy at birth (+2y)
0,0% 0,2% 0,3% 0,4% 0,4% 0,6% 0,6%
Higher Total Factor Productivity growth
(+0.4 pp)
0,0% -0,1% -0,5% -0,8% -1,0% -1,0% -1,0%
Lower Total Factor Productivity growth
(-0.4 pp)
0,0% 0,0% 0,2% 0,6% 0,7% 1,0% 0,9%
Higher employment rate of older work-
ers (+10 pp)
0,0% -0,4% -0,4% -0,4% -0,3% -0,3% -0,3%
Higher migration (+33 pp)
0,0% 0,1% -0,1% -0,1% -0,1% -0,2% -0,2%
Lower migration (-33 pp)
0,0% 0,2% 0,1% 0,2% 0,1% 0,2% 0,2%
Lower fertility (-20 pp)
0,0% 0,0% 0,2% 0,7% 1,1% 1,8% 1,8%
Temporary Shock Covid-19
0,0% 0,1% 0,0% 0,0% -0,2% -0,1% -0,1%
Permanent Shock Covid-19
0,0% 0,4% 0,7% 1,2% 1,4% 1,6% 1,6%
Linking retirement age to change in LE
0,0% -1,1% -1,9% -2,3% -2,4% -2,6% -2,6%
Unchanged retirement age
0,0% 0,4% 0,9% 1,2% 1,6% 2,2% 2,2%
Declining benefit ratio
0,0% 0,0% 0,6% 1,8% 2,7% 3,4% 3,4%
Source: Commission services ; Insee, DESTINIE model
39
3.6. Description of the changes in comparison with the 2006, 2009,
2012, 2015 and 2018 projections
Public pension expenditures as a share of GDP are projected to decrease over the projection period
(Table 18) as it is projected by French institutions (Conseil d’orientation des retraites, Institut national
de la statistique et des études économiques), but to a lesser extent. Compared to the 2018 exercise, the
new demographic (lower fertility rate, higher life expectancy) and macroeconomic (lower growth rate)
assumptions explain the revision (Graph 5.1).
Table 18 - Change in the public pension expenditure to GDP ratio and disaggregation
for consecutive projection exercises (pps of GDP)
Public pension ex-
penditure
Dependency ratio
effect
Coverage ratio ef-
fect
Benefit
ratio
effect
Labour
market
effect
Residual (incl.
interaction ef-
fect)
2006 Ageing Report (2004-
2050)
1,98 8,69 -1,79 -3,52 -0,93 -0,48
2009 Ageing Report (2007-
2060)
1,01 8,40 -2,20 -4,03 -0,51 -0,66
2012 Ageing Report (2010-
2060)
0,54 9,15 -3,53 -3,08 -1,23 -0,76
2015 Ageing Report (2013-
2060)
-2,76 6,75 -3,17 -4,73 -1,20 -0,41
2018 Ageing Report (2016-
2070)
-3,30 6,16 -2,91 -4,78 -1,43 -0,33
2021 Ageing Report (2019-
2070)
-2,17 7,08 -1,99 -5,90 -0,99 -0,36
Source: Commission services based on French projections
Explanatory note: Please note that the four components do not add up because of a residual component.The projection
horizon has been extended over consecutive Ageing Reports, limiting compatibility over time.
40
Figure 5.1 - Decomposition of the change (%) in pension expenditures to GDP between
the 2018 and the 2021 exercises (disability pensions excluded)
Source: Commission services, Insee, Destinie model, calculations: DG Trésor
Figure 5.2 - Decomposition of the change (%) in public pension expenditures to GDP
between the 2018 and the 2021 exercises - by type of pension
Source: Commission services, Insee: Destinie model, calculations: DG Trésor
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
2019 2024 2029 2034 2039 2044 2049 2054 2059 2064 2069
Dependency ratio Labour Market Coverage ratio Benefit ratio AR 2021 - AR2018
-0.4%
-0.2%
0.0%
0.2%
0.4%
0.6%
0.8%
1.0%
1.2%
1.4%
2019 2024 2029 2034 2039 2044 2049 2054 2059
% of GDP
Non-earnings related old-age pensions
Disability pensions (earnings and non-earnings related)
Survivors' pensions
Old-age and early pensions
Change in public pension expenditures between 2018 and 2021
41
The deterioration is mainly linked to the dependency ratio which has been revised upwards
compared to 2018 (Graph 5.4). The assumption of a lower fertility leads to a decrease in the projection
of the population between the age of 20 and 64 (Graph 5.3). Unlike in the previous exercise, the fertility
rate is not high enough to offset the aging of the numerous people born between 1945 and 1975. As a
result, the dependency ratio keeps raising after 2040.
Lower births can no longer compensate for the aging of the baby boomer. Consequently, the active
population keep on falling and the dependency ratio keeps on rising after 2040. This phenomenon
outweighs the evolution of other factors that contribute to containing expenditures, notably the
downward revision of the coverage ratio (retirees among those over 65) and after 2034, the downward
revision of the macroeconomic replacement rate (relative level of the average pension compared to the
average wage). The labor market indicator (ratio of the population aged 20-64 years old to the employed
population) is revised downwards due to the decline in the working age population.
Figure 5.3 - Working age population (20-64) – 2018 and 2021 projections
Source: Commission services, DG Trésor
32.000
33.000
34.000
35.000
36.000
37.000
38.000
39.000
40.000
41.000
2017
2019
2021
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
2051
2053
2055
2057
2059
2061
2063
2065
2067
2069
AR 2021 AR 2018
42
Figure 5.4 - Dependency ratio – 2021 and 2018 projections
Source: Commission services, DG Trésor
A lower projected growth rate – mainly due to lower projections for participation rates - also
contributes to a higher public pension expenditure-to-GDP ratio.
m
0
0.5
1
1.5
2
2023
2025
2027
2029
2031
2033
2035
2037
2039
2041
2043
2045
2047
2049
2051
2053
2055
2057
2059
AWG 2021 AWG 2018
43
On the contrary, the benefit ratio is lower than in the previous projection exercise after 2034 (Graph
5.1) because of a decrease of the average pension. The shorter-term benefit ratio is higher due to a
decline in the average income. The coverage ratio has been revised downwards over the whole forecast
period because of lower participation rates which lead the workers to extend their career in order to
obtain a full pension.
The increase of the non-earnings related old-age pension’s expenditures to GDP (Graph 5.2) is
linked to a higher number of pensioners which exceeds the decline in the average old-age minimum
pension.
Table 19 shows that changes in assumptions are the main drivers of the revision compared to the 2018
projections. Reforms also play a role but to a lower extent while the changes related to the new
indexation rule of the non-contributory minimum pension (ASPA) have only a small impact.
Table 19a - Disaggregation of the difference between the 2018 projections and actual
public pension expenditure in 2016-2019 (% GDP)
Source: Insee, Destinie model, c
2016 2017 2018
Ageing Report 2018 projections 15,05%
15,03% 14,98%
Assumptions (pps of GDP) 0,25% 0,17% 0,28%
Coverage of projections (pps of GDP) -0,28% -0,27% -0,20%
Constant policy impact (pps of GDP) 0,04% 0,00% -0,16%
Policy-related impact (pps of GDP) 0,00% 0,00% 0,00%
Actual public pension expenditures 15,07%
14,93% 14,90%
alculationsSource: Insee, Destinie model, calculations: DG Trésor
Table 19b - Decomposition of the difference between the 2018 and the new public
pension projections (% GDP)
2019 2030 2040 2050 2060 2070
Ageing report 2018 projections 15,01 15,45 15,08 13,76 12,54 11,75
Change in assumptions (pps GDP)
0,34 0,94 0,94 1,14 1,41 1,47
Improvement in the coverage or in
the modelling
-0,60 -0,80 -0,87 -0,66 -0,60 -0,66
Change in the interpretation of con-
stant policy
0,00 0,00 0,00 0,00 0,00 0,03
Policy related changes
0,02 0,02 0,02 0,02 0,01 0,01
New projection 14,76 15,61 15,17 14,25 13,37 12,60
Source: Insee, Destinie model, calculations: DG Trésor
44
4. Description of the pension projection model and its base data
4.1. Institutional context in which the projections are made
Several French institutions have developed pension projection models:
- Since the mid-1990s, the French statistical institute (Insee – Institut national de la statistique et
des études économiques) has developed a dynamic microsimulation model called “Destinie”.
- The Ministry of social affairs recently built up a microsimulation model called “Trajectoire”.
- The Institut des politiques publiques (IPP), a scientific partnership between the Paris school of
economics (PSE) and the Center for research in economics and statistics (Crest), has developed a
dynamic microsimulation model of the pension system called PENSIPP.
- Most pension schemes have developed their own projection model. Some of these models project
the entire pension system, like Prism created by the main private sector scheme (Cnav – Caisse
nationale d’assurance vieillesse);
- The Conseil d’orientation des retraites (COR – French pension advisory council) carries out
projections on a regular basis. The last projections
25
were published in November 2020, using
projections from all schemes.
All these projection models are often peer-reviewed, mainly during the working groups set up by the
COR.
As for the 2018 Ageing Report, the French Treasury has worked in cooperation with the French
Statistical Office using its dynamic microsimulation model, Destinie. This microsimulation model,
developed in the 90s, is a reference
26
concerning pension expenditures projections. The Destinie model
has been used for scientific studies whose results have been published in professional publications
27
28
as well as peer-reviewed journals
29
30
. It has also been used for public and officical reports
31
.
With regards to disability pensions, the projection model is the same as the one used for the 2018 Ageing
Report. This projection methodology has been developed by the French Treasury.
4.2. Data used to run the model
Old-age and survivors’ pensions projection: Destinie
The main input database is the 2018 Household Wealth Survey “Enquête Patrimoine 2017-2018”
produced by Insee. Data are collected from more than 20,000 households and provide comprehensive
25
«Evolutions et perspectives des retraites en France», 26 November 2020, COR, https://www.cor-
retraites.fr/sites/default/files/2020-12/Fusion_rapport%2Bsynth%C3%A8se_0.pdf
26
Other models like Prism (Cnav), Pensipp (IPP), or Promess (the ancestor of Trajectoire at the Ministry of social
affairs) are similar to Destinie Model.
27
Bachelet, M., A. Leduc, A. Marino, « Les biographies du modèle Destinie II : rebasage et projection », Working
paper n° G 2014/01, Direction des Etudes et Synthèses Economiques, February 2014.
28
Marino, A., « Vingt ans de réformes des retraites : quelle contribution des règles d’indexation ? », Insee
Analyses n°17, April 2014.
29
Blanchet, D., S. Buffeteau, E. Crenner and S. Le Minez, « Le modèle de microsimulation Destinie 2 : principales
caractéristiques et premiers résultats », Economie et Statistique n°441-442, October 2011.
30
Bachelet, M., M. Beffy, D. Blanchet, « Projeter l’impact des réformes des retraites sur l’activité des 55 ans et
plus : une comparaison de trois modèles », Economie et Statistique n°441-442, October 2011.
31
Rapport de la Commission Moreau pour l’avenir des retraites, « Nos retraites demain : équilibre financier et
justice », June 2013.
45
information on the household situation (professional and family biography, income and financial
situation, etc.). The model also relies on additional surveys which provide complementary information
on the labour market, or the population structure:
– Labour Force Survey (1990-2014, « Enquête emploi en continu »),
– « Échantillon interrégime de cotisants » (survey conducted by the Ministry of Social Affairs),
– Training and vocational skills survey (2015, « Enquête formation et qualification
professionnelle »).
– So called « Generation surveys » (Enquêtes generation) that focus on early carreer and transition
from school.
Disability pensions:
For disability pensions, the initial profile for recipients and average amount of the disability benefits
come from the administrative dataset of the Health insurance schemes which delivers the earning related
pensions, and from the CNAF (Caisse nationale des allocations familiales - national family insurance
fund) which delivers non earnings-related disability benefits.
4.3. General description of the model
Old-age and survivors’ pensions projection: Destinie
The Destinie model is a dynamic microsimulation model whose main application is the analysis of
pension policies and forecasting. In 2010, an updated version has been developed. This model has two
separate modules: (a) a generator of demographic and employment biographies and (b) a pension
simulator. The model takes accurately into account the household’s level and not only the individual’s
one.
(a) Biography generator
The first module produces full individual (demographic and professional) biographies (except the
transition towards retirement) up to 70 years old (or the age of death in case). Using the data from the
“Enquête Patrimoine 2010” as a starting point, the professional and family trajectories are projected
until 2070 according to transition probabilities estimated on the basis of observed data collected from
another source (see data used, 4.2).
For each individual in the sample, many variables are simulated, for instance:
– wage path estimated through wage equations (depending for instance on schooling level);
– kinship ties, which determine survivors’ pensions;
– unemployment and inactivity periods based on the estimation of transitions’ matrix on the labour
market;
– membership to different pension schemes
The sample of the Household Wealth Survey is representative of the French society with regards to:
– age and gender,
– levels of education (by generation),
– composition of households (number of children, birth/age of the mother, etc.).
– activity and unemployment rates by age and gender
Starting from the computed biographies, the model calculates the age of retirement for each individual
of the sample, assuming that people retire as soon as they meet the conditions for a full pension.
46
(b) Pension simulator
The second module is devoted to pension computation. The model is quite flexible and several
parameters can be changed: retirement behaviour, indexation of pensions, legislation scenario, etc. For
the AWG exercise, pensions have been computed according to the legislation prevailing in 2020.
Disability pensions:
The model used for disability pension projection is a macrosimulation model. It can be compared to
those used for Health Care and Long Term Care expenditure projections. The methodology is articulated
as follows:
- STEP 1: measure of the age/gender ratio of recipients and age/gender average amount of disability
benefits (ATMP pensions, PI and AAH) on the latest available dataset.
- STEP 2: calculate number of recipients for each projection year up to 2070 by multiplying the ratio of
recipients by the population by age and gender provided by Eurostat.
- STEP 3: multiply the average amount of disability benefits per age/gender on the basis of an indexation
assumption.
- STEP 4: multiply the projected average amount of disability benefits by the projected number of
recipients to obtain total projected expenditure on disability pensions.
4.4. Assumptions and methodologies applied
Old-age and survivors’ pension projection: Destinie
Sample size
The sample is composed of 65 000 individuals in 2017, with a sampling rate close to 1/1000.
Pension calculation
Since there are 35 pension schemes, Destinie covers only the main ones:
- the public sector pension scheme (FPE for civil servants in state administration, military,
CNRACL for local administration or hospitals), including the complementary part;
- the private sector pension scheme (the regime general Cnav);
- an aggregate of self-employed pension schemes (like SSI);
- an aggregate of the two point system schemes for the private sector: the complementary pension
scheme Agirc-Arrco for managers employed in private sector (Agirc), or private sector employees
(Arrco);
- one survivor’s pension scheme: this pension groups all survivors’ pension schemes, but applies
specific rules for private and public sector;
- one minimum pension scheme.
Destinie computes the first pension of the individual and makes it increase under indexation on CPI
assumption consistently with the current legislation. In general, indexation rules and parameters can be
modified by the user.
47
Survivors’ pensions calculation
Survivors’ pensions are also projected using the microsimulation model.
The Destinie model simulates the evolution of the characteristics of individuals and families, and in
particular the evolution of the marital status: separations, pairing of singles into couples, births, etc. The
model computes an individual probability of getting into a certain state, depending on the previous state
and individual characteristics.
Since the Destinie model does not distinguish between marital status, every couple is entitled to
survivors’ pensions. In real life, it is not the case: marriage provides rights for survivors’ pensions, but
not the PACS (civil solidarity pact) for instance. As a consequence, the model overestimates a little the
projections of survivors’ pensions.
The rules related to survivors’ pensions differ between pension schemes. For instance, for simplicity
reasons, the model Destinie does not split the survivor’s pension of a deceased individual between the
different former spouses or husbands he/she had (as it is the case in the main pension schemes). Other
rules specific to the public sector pension scheme, like the duration of the wedding, children, etc., are
not taken into account either.
Disability pensions
For disability pensions, the initial profile for recipients and average amount of the disability benefits
come from the administrative dataset of the Health insurance schemes which delivers the earning related
pensions, and from the CNAF (Caisse nationale des allocations familiales - national family insurance
fund) which delivers non earnings-related disability benefits.
The total amounts are recalculated using the results of the social welfare accounts. The total number of
beneficiaries is recalculated using exhaustive information published by the Ministry of Solidarity or the
organizations in charge of the payment of benefits.
4.5. Additional features of the projection model
Additional model’s characteristics (simulation of careers, simulation of the average exit age of studies
and entry age in the labour market, computation of wage equations, etc.) can be found in the 2014
professional publication (in French): Bachelet, M., A. Leduc, A. Marino, « Les biographies du modèle
Destinie II : rebasage et projection », Working paper n° G 2014/01, Direction des Etudes et Synthèses
Economiques, February 2014.
The sample of the population used to feed the Destinie model includes people living in France only, but
pensioners living abroad are included ex-post applying a constant coefficient of 1,06 as they respresent
6% of total French pensioners. Similarly, we proceed to three other ex post adjustments regarding old-
age pensions, survivors and minimum pensions to fit social security accounts balance. As for disability
pensions, no ex-post adjustments are made since the data initially used includes living abroad disability
pensioners.
48
5. Appendix
A. Methodological annex
Economy- wide average wage at retirement
The average gross wage at retirement is calculated using the average last monthly wage of new
pensioners.
Table A.1 – Economy wide average wage at retirement evolution (in EUR)
2019 2030 2040 2050 2060 2070
Economy-wide average wage 3025,15 3877,18 5368,71 7622,16 10825,72 15375,84
Economy-wide average wage at retirement 3247,64 4183,62 7165,56 9393,27 14644,08 21607,67
Source: Insee, DESTINIE model, calculations: DG Trésor
Pensioners vs Pensions
The individuals can cumulate several pension schemes depending on their careers: thus, the
number of pensioners is lower than the number of pensions.
In the model Destinie, pensioners can receive several pensions:
- Up to three defined benefit pensions (base pension scheme). In reality, there are much more
than three pension schemes but for simplification purposes only three categories are
distinguished:
• the public sector pension scheme (FPE for civil servants in state administration,
military, CNRACL for local administration or hospitals),
• the private sector pension scheme (the regime general Cnav),
• one for other pension schemes (like SSI);
- one point system schemes (complementary pension scheme), for instance, the Agirc-Arrco
for managers employed in private sector Agirc, or private sector employees Arrco. The
different point system schemes are modelized by one general point system scheme.
- one survivor’s pension. Indeed, if the deceased husband or wife had several pensions, the
surviving wife or husband may also have the corresponding survivor’s pension. We
decided to count one survivor’s pension at maximum for those individuals.
- one minimum pension.
- one disability pension. After the age of 62, earnings-related disability pensions are
considered as old-age pensions. After the age of 65 (age at which it is possible to start
receiving the non-contributory minimum pension), we suppose that individuals who used
to receive a non-earnings related disability pension also receive an old-age pension.
49
The ratio of pensions over pensioners raises from 2 in 2016 to 2.3 in 2070. This increase is due
to the fact that people are more likely to work in various sectors during their careers, which in
turn raises the probability of receiving several pensions.
Table A.1b – Pensions vs pensioners in 1000
2019 2030 2040 2050 2060 2070
Number of pensioners (I) 19029,91 21571,50 23019,50 23664,23
23882,93 23996,20
Number of pensions (II) 39302,10 46703,04 51828,99 54453,75
55503,13 55937,67
Ratio (II)/(I)
2,07 2,17 2,25 2,30 2,32 2,33
Source: Insee, Destinie model, DG Trésor
Disability pensions
There are three types of disability pensions in France. Two of them are earnings-related: the
“rente Accident du Travail et Maladie Professionnelle” (ATMP) and the “Pension d’Invalidité”
(PI). The last one, “Allocation aux Adultes Handicapés” (AAH) is a non earnings-related
minimum disability pension. In France, new disability pensions are aimed at insured individuals
under the retirement age (and only one type of disability pension (ATMP) is still being granted
after entry into retirement) so the increase of life expectancy has a limited influence on
disability pensions. As a consequence, the ratio of the number of recipients over the whole
population is supposed to be constant over time. In that sense, the projection looks like the
demography scenario of the Long Term Care methodology.
The level of new earnings-related disability pensions grows in line with the average wage. As
for non earnings-related benefits, they are price indexed.
Table A.2 – Disability rates by age groups (%)
2016 2030 2040 2050 2060 2070
-54 5% 5% 5% 5% 5% 5%
55 - 59 14% 14% 14% 14% 14% 14%
60 - 64 6% 8% 8% 8% 8% 8%
65 - 69 2% 2% 2% 2% 2% 2%
70 - 74 1% 1% 1% 1% 1% 1%
75+ 1% 1% 1% 1% 1% 1%
Source: Commission services, DG Trésor
Survivors’ pensions
The Destinie model simulates biographic situations, and in particular the evolution of the
marital statuses: separation, weddings, births, etc. One should note that the Destinie model does
not distinguish between marital statuses: marriage provides rights for survivors’ pensions, but
50
not the PACS (civil solidarity pact) for instance. Therefore, every couple is supposed to be
married, which may result in an overestimatation of the number of survivors’ pensions.
In the projections, survivors’ pension expenditures as a share of GDP decrease (from slightly
over 1.6% in 2016 to a bit more than 0.8% in 2070). There are three explanations to this trend:
- the reduction of the gap between life expectancies of women and men:
survivors’ pensions concern women for an overwhelming majority, and this reduction
of the gap between life expectancies might reduce the period of payment of
survivors’ pensions.
- the relative increase of women employment and participation rates: survivors’ pensions
are means-tested in the main basic scheme for private sector employees: due to the
evolution of women’s careers, fewer women are expected to meet the means condition
for being eligible to a survivors’ pension in the future. Moreover, as survivors’ pensions
top revenues up to a certain ceiling, women eligible to survivors’ pensions in the future
are likely to be granted a smaller amount of money on average, as their revenues are
expected to be higher on average in the future.
- the decreasing trend of marriage rate: it automatically reduces the number of people
eligible to a survivors’ pension.
Other explanations (smaller age gap between spouses, increased number of second and third
weddings, etc.) might also influence survivors’ pensions, but they are not taken into account.
Compared to the 2018 projection exercise, the way the number of recipients of a survivor
pension is calculated has been changed : in 2018, the number of pensioners included only
pensioners who didn’t receive an old-age pension. As a result, the number of pensioners of each
category could be added up to get the total number of pensioners, but these figures could not
be used to calculate the average survivor pension. In the 2018 projection, for each type of
pension, the number of pensioners is the number of pensioners receiving that type of pension:
as a result, the numbers of pensioners don’t add up. But, the average survivor pension does
represent the average additional income received by surviving spouses. These methodological
changes do not have an impact on survivor pension expenditures.
The same holds for the number of recipients of the non contributory minimum pension.
Non earnings-related minimum pensions
The ratio between non earnings-related minimum pensions and GDP increases from 0.15% in
2016 to 0.28% at the end of the 2030’s, decreases slightly to 0.2% in the following decade and
then remains stable around that level over the rest of the projection period. The number of
minimum pensions and pensioners increases on average until 2040, is relatively stable between
2040 and 2060 and increases again slightly from 2060 onwards. These variations are mostly
related to the evolution of the number of people aged 65+, which will increase at a sustained
but decreasing pace in the first half of the projection period, will not increase between 2040 and
2060 and will start increase again from 2060 onwards. Until 2050, the value of the social
51
assistance benefit is indexed to prices as per the French legislation. After 2050, the minimum
pension is indexed to wages, as agreed with the Commission.
Contributions
Alternative pension spending disagregation
Most of the time, individuals have to contribute both to a basic and complementary schemes, all of them
compulsory. Moreover, old-age insurance is organized on a socio-professional principle. It entails two
consequences: first, people tend to benefit from more than one pension (basic + complementary) and,
second, given their career, they can benefit from more than one basic pension. Therefore, focusing on
the number of pensions instead of the number of pensioners (Table 9.b) is not appropriate in the French
case because people can cumulate several pensions (cf. Annex A), which is difficult to interpret. In the
model, the coverage ratio effect is really low between 2040 and 2070, mostly due to the fact that the
average number of pensions by pensioner increases during the projection period. Indeed, people are
more likely to work in various sectors during their careers because of the expected rise in labour
mobility, which in turn rises the probability of cumulating several pensions. On the contrary, the benefit
ratio effect is even more negative because the average amount of pension is lower than the average
amount of pension by pensioner.
Table A3- Factors behind the change in public pension expenditures between 2019 and
2070 (in percentage points of GDP) – pensions
2019-30 2030-40 2040-50 2050-60 2060-70 2019-70
Public pensions to GDP
0,8 -0,4 -0,9 -0,9 -0,8 -2,2
Dependency ratio effect
3,4 2,3 0,9 0,3 0,2 7,1
Coverage ratio effect*
-1,1 -0,6 -0,1 -0,1 -0,1 -2,0
Coverage ratio old-age -0,1 -0,1 0,0 0,0 0,0 -0,2
Coverage ratio early-age -1,3 -1,7 0,0 -0,5 -0,2 -3,7
Cohort effect -3,0 -2,2 -1,1 0,0 -0,1 -6,5
Benefit ratio effect
-1,0 -1,6 -1,4 -1,1 -0,9 -5,9
Labour market effect
-0,3 -0,4 -0,2 0,0 0,0 -1,0
Employment ratio effect -0,2 -0,3 -0,2 0,0 0,0 -0,6
Labour intensity effect 0,0 0,0 0,0 0,0 0,0 0,0
Career shift effect -0,2 -0,1 -0,1 0,0 0,0 -0,4
Residual
-0,2 -0,1 0,0 0,0 0,0 -0,4
52
Administrative data on new pensioners
Tables A4 show the number of new old-age, disability, survivor and minimum Pension ASPA
pensioners by age group and sex across all pension schemes for the year 2018, produced by the
statistical office of the Ministry for Solidarity and Health, Drees. These numbers are estimated
using administrative data collected from the largest pension schemes on a yearly basis and a
survey of pensioners and demographics. An adjustment on the margin is used to reconcile the
data source.
To get those results for the year 2018, Drees considered all new pensioners who received their
first ASPA between 01/01/2018 and 12/31/2018. Therefore, only and strictly 2018 new
pensioners are counted. Those who asked earlier or even received later are not in the results.
Moreover, pensioners who deceased during the year 2018 were not counted. So were the
pensioners who were suspended of ASPA in 2018.
In 2018, most of pensioners took up their old-age pensions between age 60 and 62. The
minimum pension age is 62 and the possibilities of early-retirement will be reduced by the
increase of the contributory period needed for a full pension (cf. 1.2), so fewer retirement before
62 are expected in the near future. Men tend to retire before women, as their higher employment
rate enable them to complete the required contributory period at a younger age.
Table A4a – Number of new pensioners by age group in 2019 (administrative data) –
men
Age group All Old-age Disability Survivor
Other (including mi-
nimum)
15 - 49 5209
4789
0
420
0
50 - 54 3204
2716
0
488
0
55 - 59 19008
16770
0
2238
0
60 - 64 279610
269694
0
2435
7481
65 - 69 62067
55724
0
384
5959
70 - 74 3650
2728
0
68
854
75+ 2149
218
0
1184
747
Source: European Commission
Table A4b – Number of new pensioners by age group in 2019 (administrative data) –
women
Age group All Old-age Disability Survivor
Other (including mi-
nimum)
15 - 49 3174
884
0
2290
0
50 - 54 3813
948
0
2865
0
55 - 59 30833
13029
0
17804
0
60 - 64 274724
255881
0
11666
7177
65 - 69 86867
76978
0
3268
6621
70 - 74 5926
1818
0
2942
1166
75+ 20607
1380
0
17121
2106
Source: European Commission
53
Table A4c – Number of new pensioners by age group in 2019 (administrative data) – To-
tal
Age group All Old-age Disability Survivor
Other (including mini-
mum)
15 - 49 8383
5673
0
2710
0
50 - 54 7017
3664
0
3353
0
55 - 59 49841
29799
0
20042
0
60 - 64 554334
525575
0
14101
14658
65 - 69 148934
132702
0
3652
12580
70 - 74 9576
4546
0
3010
2020
75+ 22756
1598
0
18305
2853
Source: European Commission
B. Retirement ages of the French pension system
Generation
Minimum ages for
early pension*
Legal age
Full rate pension
age**
Before July 1st 1951
56-59 60
65
July 1
st
- Dec 31th 1951
56-60 60 + 4 months
65 + 4 months
1952
56-60 60 + 9 months
65 + 9 months
1953
56-60 61 + 2 months
66 + 2 months
1954
56-60 61 + 7 months
66 + 7 months
1955
56+4 months-60 62 months
67 months
1956
56+8 months-60 62
67
1957
57-60 62
67
1958
57+4 months-60 62
67
1959
57+8 months-60 62
67
1960 onwards
58-60 62
67
* Depending on the contribution time of the insured person (going from the reference time + 8 quarters for the
youngest retirement age, to the reference time only for the oldest retirement age) and on the age at which people
started working.
For instance, someone born in 1960 can retire at age 58 only if he/she started working at 16 and has validated
174 quarters; or at age 60 if he/she started working at 18 and has validated 166 quarters; etc.
54
C. Full pension contribution period
Generation
Required number of
contribution years
Before 1948
40 years
1949
40 years and 3 months
1950
40 years and 6 months
1951
40 years and 9 months
1952
41 years
1953 and 1954
41 years and 3 months
1955 to 1957
41 years and 6 months
1958 to 1960
41 years and 9 months
1961 to 1963
42 years
1964 to 1966
42 years and 3 months
1967 to 1969
42 years and 6 months
1970 to 1972
42 years and 9 months
1973 onwards
43 years
55
D. Pension expenditures projected by other institutions
Graph D.1 - Public pension expenditures (excluding disability) projected by other
institutions
Source: Insee (DESTINIE model) and COR projections; calculations: DG Trésor
We present a decomposition of the differences between the 2021 AWG and the 2020 COR (French
pension advisory council) exercises. The benefit ratio and the labour market ratios are slightly
different than in the country fiche decomposition, as the figures for the number of hours worked
are not available in the COR assumptions. We replaced the number of hours worked by the number
of employees
32
. The factors are the ones below:
Dependency ratio=
Population 65+
Population 20-64
Coverage ratio=
Pensioners
Population 65+
Benefit ratio=
Average pension by pensioner
Average GDP per employee
=
Pension expenditures
Number of pensioners
GDP
Number of employees
Labour market=
Population 20-64
Number of employees
32
Thus the small effect of the evolution of the number of hours worked by employee is neglected in this
decomposition.
10.0%
11.0%
12.0%
13.0%
14.0%
15.0%
16.0%
17.0%
Evolution of public pensions (%GDP)
Public pensions COR 2020 Public pensions AWG 2021
56
The decomposition of the differences in the public pension expenditures
33
between the AWG and
the COR projections is presented in Graph D.2.
Graph D.2 - Differences in the evolution between the results of the 2020 COR exercise
(based on a productivity assumption of 1.5%) and the 2021 AWG exercise (baseline
scenario excluding disability pensions)
Source: Insee (DESTINIE model) and COR projections; calculations: DG Trésor
In this graph we study the difference between the gap of each year AWG projection to 2018 AWG
projection, and the gap of each year COR projection to 2018 COR projection.
One of the main differences between the two projection exercises stems from demographic
assumptions: the Eurostat assumptions has changed the projections of life expectancy and fertility.
Based on these assumptions, the dependency ratio will continue to increase until 2045 before
declining until 2070.
Over the medium term, differences between the coverage ratios also explain the divergent path of
the two projections. These differences come mostly from the fact that the modeling techniques
used by the COR and the French Statistical Office are not the same: the COR aggregates
projections made by the statistical services of each scheme while the French Statistical Office
makes projections based on a sample of households and observed data. The evolution of the benefit
ratio diverges between the two exercises during the projection period, which is due to lower
productivity growth rate assumptions in the AWG exercise.
At the beginning of the projection period, both AWG and COR pension expenditures increase with
a slightly higher jump observed regarding COR expenditure, since the decline in GDP modelled
by the COR linked to the Covid-19 crisis is slightly larger than for the AWG scenario. From 2022
onwards we observe higher public pension expenditures in the AWG scenario when compared to
33
To be as consistent as possible with the COR projections’ field, disability pensions have been excluded, but
minimum pensions are included as they are also taken into account in the COR projections.
-3.00%
-2.00%
-1.00%
0.00%
1.00%
2.00%
3.00%
2018
2020
2022
2024
2026
2028
2030
2032
2034
2036
2038
2040
2042
2044
2046
2048
2050
2052
2054
2056
2058
2060
2062
2064
2066
2068
2070
Evolution AWG2021 - COR2020
Dependency Ratio Benefit Ratio Coverage Ratio
Labour Market Public pensions to GDP
57
the COR, before a reduction of the gap and a reversion by 2060. This evolution is related to several
factors:
• Fertility and Total Factor Productivity (TFP) growth assumptions are higher and conse-
quently more advantageous in the COR scenario, implying lower expenditures when com-
pared to the AWG scenario. However, the gap between the COR and AWG TFP growth
assumptions shrinks over the projection period, implying a decline in expenditures smaller
over time when compared to the AWG scenario.
• This decline in expenditures, which is smaller over time, is outweighed by a higher life
expectancy and a lower migration in the COR scenario, with both leading to an increase in
the COR expenditures relative to the AWG projections.
• Furthermore, we observe that employment rate of older workers is higher in the COR sce-
nario when compared to AWG one up to 2055 before dropping below the AWG's employ-
ment rates, which implies lower pension expenditures in the COR scenario up to 2055 and
then an increase in expenditures from 2055 onwards regarding COR scenario.
58
E. Projected and disaggregated new public pension expenditure
(old-age and early earnings-related pensions)
Table F.1 – Decomposition of new pension expenditures: computation of the main variables
34
With this definition:
=
×
×
×
.
New pension expenditures P
=
∑
where
is the annual pension of
the new pensioner provided by Destinie.
Number of new pensioners N
Provided by Destinie.
Average contributory period (in years)
̅
=
∑
where
is the number of years
of a positive wage for the new pensioner
(whose complete wage series is provided by
Destinie).
Average number of months paid the first
year
=
∑
where
is the number of
months of pension paid to the new pensioner
the first year (provided by Destinie).
Defined benefits schemes
Monthly average pensionable earning
Computed using the 25 best year wages (series
provided by Destinie) as
=
∑
where
=
∑
Ι
,
,
(
1+
)
and
is
the CPI and Ι
,
=1 if
,
is one of the 25
best yearly wages of the individual .
Average accrual rate
Computed so as to resolve
=×
̅
×
×
× .
Thus is close but not equal to =
∑
where
is defined by:
=
∑
,
(
1+
)
34
.
Point system schemes
Total pensions points at retirement
̅
=
∑
where
is the number of pen-
sions points acquired by new pensioner at re-
tirement (provided by Destinie, Agirc points
are converted into Arrco points using the re-
spective points value in both schemes).
Point value
Service value in the Arcco scheme
Point cost
Purchase value in the Arcco scheme multiplied
by the adjustment factor applied to contribu-
tions (125% then 127% after 2019)
Adjustment factor
Computed so as to resolve
=×
̅
×
×
×
̅
.
Thus
̅
is close but not equal to 1.
59
Table F.2 - Projected and disaggregated new public pension expenditure (old-age and early
earnings-related pensions) - Defined-benefit schemes
Source: Insee, DESTINIE model, calculations: DG Trésor
Table F.3 - Projected and disaggregated new public pension expenditure (old-age and early
earnings-related pensions) - Point systems
2019 2030 2040 2050 2060 2070
Projected new pension expenditure (mln € ) 1816,4 2532,6 4682,6 7024,5 8564,5 11610,7
I. Number of new pensioners ('000) 556,3 707,2 707,2 756,3 668,9 694,0
II. Point value (EUR/month) 0,005 0,005 0,005 0,005 0,005 0,005
III. Average accrual rate 4221,5 4795,5 6751,2 8547,8 11736,4 15295,5
IV. Total pension points at retirement (by
pension)
108119,4 106399,1 160987,9 210321,2 288267,6 370897,9
V. Average contributory period 25,612 22,2 23,8 24,6 24,6 24,2
VI. Sustainability/Adjustment factor 1,000 1,0 1,0 1,0 1,0 1,0
VII. Correction coefficient 0,931 0,9 1,0 1,1 1,1 1,0
VII. Average number of months paid the first
year
5,921 6,66 7,66 8,02 8,14 8,29
Source: Insee, DESTINIE model, calculations: DG Trésor
2019 2030 2040 2050 2060 2070
Projected new pension expenditures (million EUR)* 4014 7481 11 601 16 972 20 621 31 831
I. Number of new pensioners ('000) 621,9 804,4 772,7 803,3 698,4 737,8
II. Average contributory period 33,0 31,1 32,8 32,7 32,8 33,0
III. Average accrual rate (%) 1,1 1,2 1,1 1,1 1,0 1,0
IV. Monthly average pensionable earnings (€ ) 3 077 3 633 5 5848 7 490 10 656 15 244
V. Sustainability/Adjustment factor 1 1 1 1 1 1
VI. Average number of months paid the first year 5,9 6,6 7,6 8,0 8,1 8,2
(Monthly average pensionable earnings) /
(Monthly economy-wide average wage)
101,7% 93,7% 104,0% 98,3% 98,4% 99,1%
60
F. Overview of the sensitivity tests
Sensitivity test Definition
1 Higher life expectancy
Increase of life expectancy at birth of two years by 2070 compared with the
baseline projection.
2
Higher total factor
productivity growth
Total factor productivity growth is assumed to converge by 2045 to a steady-
state growth rate which is 0.2 percentage points higher than in the baseline
scenario. The increase is introduced linearly during the period 2026-2045
3
Lower total factor
productivity growth
Total factor productivity growth is assumed to converge by 2045 to a steady-
state growth rate which is 0.2 percentage points lower than in the baseline
scenario. The increase is introduced linearly during the period 2026-2045
4
Higher employment
rate of older workers
The employment rate is 10 p.p. higher compared with the baseline projection
for the age-group 55-74. The increase is introduced linearly over the period
2021-2030 and remains 10 p.p. higher thereafter. The higher employment
rate of this group is achieved through a reduction of the inactive population.
5 Higher migration
A scenario whereby net migration flows are 33% higher than in the baseline
scenario over the entire projection horizon
6 Lower migration
A scenario whereby net migration flows are 33% lower than in the baseline
scenario over the entire projection horizon
7 Low fertility
A scenario whereby net fertility decreases by 20% over the entire projection
horizon
8
Linking retirement
age
The retirement age is shifted year-over-year in line with ¾ of the change in
life expectancy at current retirement ages.
9
Unchanged retirement
age
The early and statutory retirement ages, as well as career length require-
ments, are frozen at the situation in the base year.
10 Declining benefit ratio
When the earnings-related public pension benefit ratio declines by more than
10% as compared to the base year level, pension indexation is increased to
stabilise the benefit ratio.
11 Permanent shock
The impact of the Covid 19 crisis is modelled as permanent shock
12 Temporary shock
The impact of the Covid 19 crisis is modelled as temporary shock
61
Graph G.1 – Pension expenditures under various scenarios (% of GDP)
Graph G.2 – Pension expenditures under various scenarios (% of GDP)
11.00%
12.00%
13.00%
14.00%
15.00%
16.00%
HLE Higher TFP Lower TFP
Higher Employement Higher Migra Lower Migra
Baseline
9.50%
10.50%
11.50%
12.50%
13.50%
14.50%
15.50%
16.50%
17.50%
Baseline Lower fertility Temporary shock
Permanent shock Link retirement to LE Unchanged retirement rate
Declining benefit ratio
62
G. Panorama of the main pension schemes
Source: GIP info retraite, www.info-retraite.fr
63
H. Glossary
CNAV: Caisse nationale d’assurance vieilesse
AGIRC: General association of complementary retirement institutions for executives /
Association générale des institutions de retraite complémentaire des cadres
ARRCO: Association pour le Régime de Retraite Complémentaire des salariés
IRCANTEC: Institution de retraite complémentaire des agents non titulaires de l’État et des
collectivités publiques
MSA: Mutualité Sociale Agricole
FPE: Fonction publique d’Etat
RAFP: Retraite additionnelle de la fonction publique
CNRACL: Caisse Nationale de Retraités des Agents des Collectivités Locales
RATP: Régie autonome des transports parisiens
SNCF: Société nationale des chemins de fer
CNIEG: Caisse nationale des industries electriques et gazières
SSI: Sécurité sociale independants
CNAVPL: Caisse nationale d’assurance vieilesse professions liberales
CNBF: Caisse nationale des bareaux français
RCI: Retraite complémentaire independants
CAVP: Caisse d'assurance vieillesse des pharmaciens
CARCDSF: Caisse Autonome de Retraite des Chirurgiens Dentiste et des Sages Femmes
CARPIMKO: la caisse autonome de retraite et de prévoyance des infirmiers, masseurs-
kinésithérapeutes, pédicures-podologues, orthophonistes et orthoptistes
CARPV: Caisse Autonome de Retraite et de Prévoyance des Vétérinaires
CAVEC: Caisse d'assurance vieillesse des experts-comptables et des commissaires aux compte
CAVAMAC: Caisse d'Allocation Vieillesse des Agents Généraux d’assurance
CRN: Caisse de retraite des notaires
CAVOM: Caisse d’assurance vieillesse des officiers ministeriels, des officiers publics et des
compagnies judiciaires
CIPAV: Caisse interprofessionnelle de prévoyance et d’assurance vieillesse
FPT : Fonction Publique Territoriale
FPH : Fonction Publique Hospitalière
64
6. References
• « Évolutions et perspectives des retraites en France», Conseil d’Orientation des Retraites, June
2017, Yearly Report.
• Rapport de la Commission Moreau pour l’avenir des retraites, « Nos retraites demain : équilibre
financier et justice », June 2013.
• « Les biographies du modèle Destinie II : rebasage et projection », Working paper
n° G 2014/01, M. Bachelet, A. Leduc, A. Marino, Direction des Etudes et Synthèses
Economiques, February 2014
• « Vingt ans de réformes des retraites : quelle contribution des règles d’indexation ? »,
A. Marino, Insee Analyses n°17, April 2014.
• « Le modèle de microsimulation Destinie 2 : principales caractéristiques et premiers résultats
», D. Blanchet, S. Buffeteau, E. Crenner and S. Le Minez, Economie et Statistique n°441-442,
October 2011
• « Projeter l’impact des réformes des retraites sur l’activité des 55 ans et plus : une comparaison
de trois modèles », M. Bachelet, M. Beffy, D. Blanchet, Economie et Statistique n°441-442,
October 2011
• « Modulations de la retraite selon l’âge de départ : principes directeurs et évolutions depuis les
années 1980 », K. Briard and S. Mahfouz, Economie et Statistique n°441-442, October 2011.
• « The October 2015 agreement on France's complementary pension schemes for private-sector
employees (AGIRC and ARRCO) will improve the pension system balance by 0.3 points of
GDP from 2020 to 2060 », Julia Cuvilliez and Thomas Laurent, Direction générale du Trésor,
Trésor-economics n°178, August 2016.
• « French retirement benefit expenditure set to shrink substantially as a share of GDP by 2060,
according to European projections », Julia Cuvilliez, Geoffrey Lefebvre, Pierre Lissot, Yves
Dubois and Malik Koubi, Direction générale du Trésor and Insee, Trésor-Economics
n°152/Insee Analyses n°21, August 2015.
