1In France, social protection expenditure represents almost 672 billion euros or 31.8% of GDP. Within the OECD, the country distinguishes itself by the high public share of social protection expenditure and by the fact that the social protection system is divided between various different institutions. This fragmented governance raises three fundamental economic problems. First, it makes it more difficult to take collective decisions regarding the size of public expenditure on social protection and its distribution by risks. Second, the lack of coordination between basic and complementary protection schemes, regarding to any given social risk, does not ensure that needs are met at the lowest societal cost. Third, the present organisation prevents a clear distinction between two types of social protection that differ in their approaches to solidarity and financing: schemes that pay out contributory benefits with the objective to provide a substitute for earned income, and social protection systems offering benefits to all citizens according to their needs, unrelated to the amount of contributions.
2Although these facts are widely acknowledged, there is little consensus on proposals for reform. We propose an ambitious reform of the social protection architecture, which could be envisaged in the medium to long-term. In the first place, this consists of unifying the schemes that cover the same risk: unifying the governance of compulsory pension schemes and establishing consistent rules for the accrual of entitlements; channelling health risk cover to the basket of social healthcare items and refocusing optional insurance on healthcare that falls outside of this basket. Moreover, the architecture of social protection should be reorganised with a non-contributory section (family, health insurance, fight against poverty, etc.) incorporated into the State budget and financed by taxation, and a contributory section (retirement pensions, unemployment insurance, daily allowances of sickness benefit, etc.) financed by social security contributions. In the shorter term, measures are recommended in order to improve the management of old-age insurance: the dependence of the balance of the system upon long-term economic growth could be reduced by re-establishing index-linking to wages, constituting the reference wage for pensions, and adjusting the income replacement rate by using a demographic coefficient. In order to guarantee real healthcare insurance cover, public cover should be established for 100% of remaining payable costs for recognised expenditure, beyond a certain threshold, while guaranteeing access to healthcare provision at price rates that are valid throughout French territory.
3Finally, for the sake of informed choices, every year at the time of the budgetary debate an annex should be presented setting out a breakdown of public expenditure as a whole, according to function and social risk covered.
4Social protection expenditure in France currently accounts for nearly 672 billion euros, i.e 31.8% of GDP.  The vast majority of this sum comprises public expenditure (609 billion euros or 28.8% of GDP). This expenditure is at the heart of the “French model” of social protection and is an essential part of the debate on sound public finances. In a context of major budgetary constraints, a sudden increase of social expenditure exceeding national income growth requires tighter management to cover needs at the lowest cost, while acting in accordance with citizens’ preferences.
5The slow construction process of the French social protection system has led to a complex architecture, with a large number of different actors taking uncoordinated decisions, and leading in the end to a lack of transparency in decision-making. This Note du CAE aims to review this architecture and suggest courses of action for reorganisation enabling both the expression of democratic choices and the indispensable search for effectiveness, with the focus on retirement pensions and health insurance, which represent almost 80% of social protection expenditure.
The field of social protection in France
6There is no universal definition of the scope of social protection. Since institutional arrangements vary from country to country, it is difficult to distinguish the boundaries of social protection. Eurostat defines it as encompassing “all interventions from public or private bodies intended to relieve households and individuals of the burden of a defined set of risks or needs, provided that there is neither a simultaneous reciprocal nor an individual arrangement involved” (ESSPROS Manual).  In practice, social protection accounts include total expenditure aimed at covering sickness, maternity, disablement, industrial accidents, old age, survivors, family, unemployment, professional integration and reintegration, housing, poverty and social exclusion risks, whatever the schemes involved.  Within the 672 billion euros of social protection expenditure for 2013, old-age/survivors insurance accounted for 307 billion euros while sickness-maternity-disablement insurance accounted for 228 million euros (Graph 1).
Breakdown of social protection expenditure, in billions of euros
Breakdown of social protection expenditure, in billions of eurosNote: a Poverty-social exclusion (16.8) + Housing (17.7) + Professional integration-reintegration (3.4) + Workplace accidents (12.2).
- With regard to total social protection expenditure (without distinction between public and private expenditure) France is among the highest-spending countries, although the differences between countries remain limited;
- France distinguishes itself primarily by a higher proportion of public expenditure on retirement pensions.
8Our country also stands out by the fact that its social protection system is divided between a large number of different institutions.  The general national social security system, created in 1945, is at the heart of the current system, providing sickness, disablement, old age, industrial accident and family cover for the majority of private sector employees.
9However, with the exception of the family branch, the “general” social security system has never had the universality that its designers desired, due to the fact that state employees (already well-covered), the self-employed (who preferred their own specific schemes) and employees of large public-sector firms (the “special schemes”) were excluded from the outset.
10Moreover, important components of social protection were left out of the social security system and taken care of by other institutions. The absence of unemployment insurance lead the social partners to create the UNEDIC unemployment benefits agency at the end of 1958, an insurance scheme outside of the scope of the social security system, which was made compulsory the following year. The low level of old age pensions also led the social partners to create a complementary pension scheme for managerial staff (the AGIRC) in 1947 and its equivalent for non-managerial staff (the ARRCO) in 1961: originally optional, they became compulsory in 1972. As far as the healthcare branch is concerned, complementary health insurance developed in order to cover the remaining payable healthcare expenditure costs not covered by the general social security system. Today, this remains outside of the field of public expenditure, although it plays an important role in health coverage in France (cf. Note CAE no 12). 
Social protection expenditure: management in question
The division between contributory and non-contributory financing remains incomplete
11The social security system was originally designed based on the social insurance model: a close relationship between contributions and benefits was established and affiliation was connected with the exercise of a professional activity (although coverage was automatically extended to the family of the insured person).
12The universalisation of health-insurance benefits and family benefits, the major increase in the content and costs of healthcare services covered by the social security system, as well as the removal of caps on social security contributions progressively weakened the connection between contributions and benefits. Moreover, the policy of reductions in employers’ social security contributions on low wages and the introduction of solidarity mechanisms (social minima, housing benefits, etc.) led the State to take an increasingly large share in the financing of social protection.
13These long-term developments tended towards a distinction between social security schemes paying essentially contributory benefits, aimed at providing a substitute for earned income in case of loss of wages, and a system of social protection aimed at all citizens offering benefits according to needs, and without conditions regarding past contributions.  In this respect, the creation of the CSG (contribution sociale généralisée, supplementary social security contribution) in 1991, a proportional tax on income earmarked for social protection as a replacement for social security contributions, marked a decisive stage.
14However, the clarification has not been fully completed: each scheme continues to be financed by a mixture of contributions and taxes, which contributes to the lack of transparency in compulsory deductions for the financing of social protection. Yet, without going into the reflections of political economy concerning social protection, economic theory suggests that greater effectiveness of compulsory levies can be obtained by financing contributory benefits by social security contributions. This is due to the fact that social security contributions constitute a fixed-rate levy and establish a direct relation to the level of benefit individuals have in return. In this extent, certain results from the literature show a lower prevalence of avoidance behaviours.  In more specific terms with regard to the French case, which is remarkable for the generosity of its retirement pension system (basic + complementary), the level of compulsory levies will only be tenable over time if the visible connection between deductions and benefits is maintained.
15For non-contributory benefits such as healthcare, which are provided in proportion to need, there are no disadvantages in establishing financing through taxation. In France, revenue assignation is the general rule. However, economic literature is very critical of earmarked taxes,  since earmarking creates constraints that are harmful to the optimal distribution of public financing. The rigidities thus created are illustrated by the obstacles to plans such as the incorporation of reduced social security contributions into the scale, or the rationalisation of numerous compulsory levies, partly (but not exclusively) prevented by the institutional mixture of contributions-based and universal entitlements. 
16The following table presents schematically the division of revenue and items of expenditure according to their contributory or non-contributory character. Apart from the complexity of social protection financing, there is little difference in the division between contributory and non-contributory financing regarding receipts and expenditure. With financial costs set aside, contributory expenditure (proportional to insured persons’ incomes) represents around 54% of total expenditure, whereas social contributions account for 62% of resources. For this reason, clarification of the financing of social protection could be envisaged without disrupting the broad financial balances.
Social protection resources and expenditure according to contributory and non-contributory criteria
Social protection resources and expenditure according to contributory and non-contributory criteriaField: Expenditure of general government and compulsory levies.
Interpretation: Non-contributory old age and unemployment items of expenditure (social minima, family and pensions entitlements) are included under the fight against poverty and social exclusion or the family.
17This question is well-known to social protection specialists, and has been the subject of very comprehensive reports from the Haut conseil du financement de la protection sociale (HCFi-PS, French High Council for the Funding of Social Protection).  The HCFi-PS explores several scenarios transferring social security contributions from “universal” branches (such as the family branch) to “contributory” branches (the old age branch) and, conversely, the financing of “universal” branches through taxation. The objective is to reserve social security contributions to principally contributory schemes. The scenarios put forward all have technical difficulties, although they don’t appear to be overwhelming.
Governance and and collective trade-off between risks
18Social protection expenditure was long neglected by the budgetary control process. Indeed, since social security administrations are not included in the State budget, increasing social protection expenditure does not fall within the field of Budget Acts (LF / Lois des finances). In order to remedy this situation, the constitutional reform of February 1996 increased the Parliament’s authority regarding social security finances, by instituting the vote of a Social Security Finance Act (loi de financement de la Sécurité sociale, LFSS) every year. Social Security Finance Acts (LFSS) contain spending targets for each branch, in addition to a National Health Insurance Spending Objective (Objectif national de dépenses d’assurance maladie, ONDAM) covering the compulsory basic schemes as a whole. Each LFSS covers all schemes within the social security system, reaching from general to special schemes, but not the non-compulsory mechanisms provided by institutions outside of general government (complementary health insurance, provident insurance, etc.) nor compulsory regimes resulting from equi-representational union-employer negotiation (complementary retirement pension schemes, UNEDIC). In total, 24% of public social protection expenditure is covered neither by Social Security Finance Acts nor by Budget Acts (Graph 2),  whereas in most European countries compulsory public expenditure is included in a single budgetary text subject to Parliamentary control.
Governance of social protection expenditure, in % of total expenditure
Governance of social protection expenditure, in % of total expenditureInterpretation: 64 % des dépenses de protection sociale en 2013 sont couvertes par le PLFSS, et 17 % sont déterminées par les partenaires sociaux.
19The current governance poses two fundamental economic problems. The first concerns the difficulty of achieving collective trade-offs on the basis of fragmented governance. The democratic process currently in place, whether parliamentary or based upon equi-representational union-employer negotiation, does not guarantee that choices between different items of social protection expenditure are made in accordance with the expression of collective choices.
20The second problem involves the poor coordination between basic and complementary schemes covering the same risk. This point is examined below regarding retirement pensions and healthcare insurance.
Incomplete Control of Risk Coverage
22In France, 78% of health expenditure (consumption of medical care and materials) is covered by the social security system and the State, 13.5% by complementary insurance bodies, with the remainder, i.e. 8.5%, being financed by households through direct payments.  Within the framework of the Social Security Finance Acts (LFSS), the National Health Insurance Spending Objectives (ONDAM) voted by Parliament cover 78 % of health expenditure. In order to guide Members of Parliament, Social Security Finance Bills (PLFSS) provide facts and figures on developments regarding the coverage rate of the national social security system and the State (field of the ONDAM). They also indicate the coverage rate of complementary insurance bodies, as well as the proportion of direct payments made by households (remaining payable costs). All these elements are given as a proportion of consumption of medical care and materials, a very broad aggregate encompassing non-reimbursable expenditure such as private doctors and treatments not classified as reimbursable and non-reimbursable medicines. Concerning healthcare, a distinction should be made between recognised expenditure, which is equal to expenditure submitted for reimbursement within the valid price rates limit and regulatory classifications of healthcare types, and reimbursable expenditure, which includes excess fees and freedom to set price rates (for medical materials) and may be partly or wholly covered by complementary insurance schemes. Since public expenditure is intended to guarantee equal access to basic healthcare, for which solidarity mechanisms have been established (connected to a reference basket, which might be referred to as a “basket of social healthcare items”),  it would be appropriate to provide further information to Parliament concerning changes in the public coverage rate, not only as a proportion of consumption of medical care and materials, but also with regard to reimbursable expenditure and recognised expenditure respectively.
23Tensions regarding public expenditure on health insurance arise from the fact that in France, as in all developed countries, health expenditure is growing more rapidly than other consumption items. This is less a matter of population ageing than of medical innovations, which are a source of progress in health and longevity. In order to control expenditure, adjustment is too often made by a reduction of reimbursement rates. Although this enables reduction of public expenditure strictly speaking, restriction of growth in National Health Insurance Spending Objectives (ONDAM) of this kind does not enable control of health expenditure as a proportion of household budgets, whose spending power is passively crippled by the supplementary part of their healthcare expenditure rather than being released. Furthermore, reduction of reimbursement rates as a tool of budgetary regulation removes responsibility from the authorities in charge of healthcare system management. Reimbursable expenditure as a whole needs to be controlled, rather than reimbursed expenditure alone. It should be emphasised that this control method of public health expenditure was abandoned in 2012, which has automatically led to an annual increase in the average coverage rate provided by the social security system. This development is principally attributable to the increased proportion of insured persons entitled to 100% cover for long-term illnesses (ALD, affections de longue durée, cf. infra),  which is itself related to population ageing. This shows the importance of putting relevant mechanisms in place in order to ensure compliance with National Health Insurance Spending Objectives (ONDAM), which in future years will necessarily require a far-reaching reorganisation of community medicine (Box 1). 
1. How to ensure long-term compliance with the ONDAM?
As emphasised by the French National Health Insurance Fund for Employees (Caisse nationale de l’assurance maladie des travailleurs salariés, CNAMTS), [a] medicines cannot be indefinitely used as a source of savings. Moreover, it will eventually become ineffective to do so, since numerous medical innovations are expected in the field of medicines. Compliance with the ONDAM will therefore require control of ambulatory care expenditure. Solutions may be found in the allocation of budgets as a whole – ambulatory, hospital and social-health care– the set-up of payments and the registration of health professionals with the Department of Health (cf. Note du CAE no 8) [b]. Moderate proposals in this direction were contained in the initial bill for the modernisation of the healthcare system (projet initial de la loi de modernisation du système de santé); they were rejected by independent professionals. In the absence of real reform of the organisation of healthcare provision, within a framework that continued to be organised on a fee-for-service basis (a source of ineffectiveness described in Note du CAE no 8), one solution for the control of ambulatory care expenditure would be to assign a floating point to elements of remuneration in order to ensure compliance with the closed envelope defined by the ONDAM. In order to avoid the potentially harmful effects of the floating point (encouraging large numbers of medical acts), a graduated price rate, after the German model, could be applied according to an individual threshold level of activity established for each general practitioner.
Compulsory and complementary Insurance
24Subscribing to complementary health insurance is optional as far as individual contracts are concerned and considered optional for group contracts. However, the large proportion of not reimbursed costs by the social-security system makes complementary insurance crucial for healthcare access in France, which means that the majority of individuals are covered: in 2012, only 5% of persons with social security healthcare cover declared that they did not have complementary insurance cover. 
25This situation creates confusion in the public debate on the scope of cover. Since complementary insurance is optional and persons without complementary cover in general have low incomes (just above the CMU-C complementary universal healthcare cover ceiling, which is lower than 75% of the poverty line), it is pertinent to assess quality of cover on the basis of remaining payable costs measured before complementary insurance is taken into account.  The Haut conseil pour l’avenir de l’assurance maladie (HCAAM, High Council for the future of the health insurance system) thus shows that, for the highest percentile in terms of consumption of medical care, the average remaining payable costs amount to almost 5,000 euros per year. A result of this order suggests that the social security system does not provide adequate protection against “catastrophic” expenses and may lead to the recommendation that capping of remaining payable costs should be established (cf. Note du CAE no 12, op. cit.).
26Another source of confusion arises from the fact that complementary insurance bodies combine the roles of complementary (insofar as they contribute to covering the basket of social healthcare items) and supplementary insurance (they also cover other care such as comfort in hospital, alternative medicine and excess fees). The fact that optional insurance plays a major role in complementing basic healthcare cover is specific to France, optional insurance generally being restricted to the role of supplementary insurance in other European countries.
27It is important to distinguish between the roles of complementary and supplementary insurance, since its regulation needs to be governed by different principles. Because they play a role of co-insurance for basic healthcare, complementary insurance providers need to define their financing and benefits in accordance with solidarity principles. Conversely, since supplementary insurance providers play a role in access to comfort care, it is entirely admissible for them to be free to determine their benefits and price rates. However, the French reality diverges from the application of these principles since the roles of complementary and supplementary insurance are inextricably intertwined in the policies provided by complementary insurance bodies.
28Radically different approaches to solidarity can thus be observed between the complementary insurance sector and the social security system. In the case of the latter, contributions are proportional to individuals’ incomes, without consideration of their age and state of health. Moreover, the long-term illnesses (ALD, affections de longue durée) system provides better cover for persons suffering from chronic illnesses. This two-tier cover is equivalent to long-term cover: persons entitled to healthcare cover not suffering from long-term illnesses agree to contribute as much as persons suffering from long-term illnesses, since they know that they will be well-covered should they ever contract a chronic illness. The frequency of chronic illnesses increasing with age, this system amounts to solidarity between young and old people.
29Conversely, current developments in the complementary insurance sector lead to the pooling of risks by age group. Indeed, private companies in competition on an optional insurance market can only provide limited pooling of risks. On the personal complementary insurance market, this results in the equivalent of age-based price rates, and very expensive insurance policies for elderly people. The obligation henceforth imposed upon private sector companies to insure all of their employees will lead to a further increase of this segmentation by excluding employees from the pooling of personal complementary insurance.
30In order to limit the likely rise in the price of personal complementary insurance policies accessible to pensioners, an article of the 2016 Social Security Finance Act (LFSS) establishes official labelling of insurance contracts in order to organise competition on the complementary insurance market for persons of 65 years of age and over. This organisation is entirely contrary to the national health insurance system’s solidarity between different age groups. Conflicting solidarity principles are thus applied by two different actors –basic and complementary insurance bodies– which nonetheless contribute to the coverage of the same type of treatments.
Contingency management of old age
A fragmented system
31There are about thirty different retirement pension systems in France. They are divided along professional lines (in particular among the self-employed and companies with special schemes), as well as on the basis of the distinction between basic and complementary schemes. As a result, many retired people have accumulated entitlements under several different compulsory pension schemes: a pensioner receives on average 2.8 pensions, 25% of them receive four or more retirement pensions.  Fragmentation of this kind might be theoretically justified in several different ways.
32First, since old-age insurance involves the striking of a fundamental balance between leisure and work, it might be deemed desirable for individuals to have freedom of choice, according to their preference, the type of work that they perform or their state of health, with regard to the proportion of their income devoted to financing their retirement pensions. This results in various different options, which enable justification of the existence of several different schemes. This is the principal reason why most of our neighbouring countries have developed company pension schemes and individual savings mechanisms, subscribed directly by employees or offered by employers. France did not follow this path, to a large extent because complementary pension schemes progressively opted for pay-as-you-go financing, giving them a compulsory character.
33Moreover, the existence of basic and supplementary schemes could be aimed at meeting two distinct objectives: a safety net for workers as a whole and a contributory pension scheme, depending upon the choices of employees and their employers. This corresponds in part to the so-called “three pillars” model defended by the World Bank (World Bank, 1994):  a first pillar providing a minimum pension, a second compulsory contributory pillar and a third personal retirement savings pillar. From this point of view, there is a profound difference between basic and supplementary schemes, both in terms of their method of financing (capitalised pension funds for supplementary schemes) and regarding their governance.
34Finally, the argument of regulated competition could be raised in favour of a large number of different pension schemes, amongst which employees (or employers) would be able to choose, according to their respective management performance. On the basis of rights and duties defined in a centralised manner, competition would potentially improve the effectiveness of management –where efficiency gains are greater than competitive costs (advertising, information, etc.).
35In the French case, it is difficult to use these arguments in order to justify the large number of different pension schemes, and particularly the two level system consisting of a basic and a complementary scheme. When the complementary schemes were made compulsory in 1972, their nature changed as they started forming a level of the legal compulsory pension scheme based on a pay-as-you-go principle. Thus, this cannot be viewed as an expression of heterogeneous personal preferences or as the establishment of competition between different schemes. Nor does the distinction between basic and complementary schemes correspond to the distributional and contributory aspects of old-age insurance. Indeed, redistribution within old age expenditure is for the most part carried out by non-contributory mechanisms (minimum old-age pension, family entitlements, inclusion of periods of unemployment, etc.), rather than the calculation difference of pensions between basic and complementary schemes.
36While the large number of different compulsory pension schemes does not provide any obvious advantage within the French institutional architecture, it however translates into a lack of coordination in between schemes, giving rise to additional administrative costs, inconsistencies and opacity of pension entitlements for insured persons (Box 2).
2. The inconsistency and opacity of pension schemes in France
The lack of coordination between different pension schemes also leads to difficulties in the management of pensions expenditure, in particular for multi-scheme systems. The case of the contributory minimum provides an enlightening example: the latter was put in place in 1983 in order to guarantee a pension higher than the minimum old-age pension for employees having contributed for long periods. However, it was not realised until 2007, thanks to a study undertaken by the DREES (French department for research, studies, assessments and statistics), that a large proportion of beneficiaries of the contributory minimum were poly-pensioners, enjoying an overall pension markedly higher than the incomes originally targeted. [b]
For private sector employees, the differences in rules between the general pension system and complementary pension schemes have always been a source of complications and incomprehension. The decision of 30th October 2015 on the part of the AGIRC-ARRCO schemes (respectively the complementary pension scheme for managerial staff and the equivalent for non-managerial staff) to introduce temporary reductions and to refer to different required lengths of contribution in complementary schemes than the general national pension system, provides a good example of absence of coordination. The effectiveness of measures on the part of complementary schemes will be limited by their exclusive application to a fraction of pensions and the system as a who le will thereby be rendered still more complex.
The case of demographic offsetting between pension schemes –mechanisms for transfers between schemes in order to correct demographic imbalances between them– provides another illustration of the difficulties posed by the fragmentation of compulsory pension schemes, within the framework of pay-as-you-go financing. Indeed, there is no reason to predict that the professional division of different pension schemes should lead to parallel patterns of change in the number of contributors and pensioners for the schemes as a whole. Quite the reverse, it is natural to expect distinct patterns of change in the numbers of farmers, miners, self-employed persons and members of other professions. In countries with a single compulsory public pension scheme (e.g. the United States, the Netherlands, the United Kingdom and Sweden), the question of demographic offsetting does not arise since it is effected in a de facto manner within the country’s professions as a whole. In France, demographic offsetting has been put in place at the level of basic pension schemes, and for direct entitlements. Complementary regimes were not included and will therefore have to manage their potential future imbalances through the accumulation of financial reserves. The great complexity of current offsetting rules and their imperfect character [c] makes it impossible to correct the incongruity of the existence of compulsory pension schemes, which are autonomous but financed on a pay-as-you-go basis.
37Due to the complexity of the law, overlapping rules and lack of management tools available for insured persons to simulate their actual replacement rate and the amount of their total pension based on their age of departure from employment, etc., contributors are faced with a particularly opaque system. Important progress has been made, through the right to information, with the creation of a public interest grouping (GIP, groupement d’intérêt public) Info Retraite and, today, its transformation into GIP Union Retraite, but these commendable initiatives have not reduced institutional complexity. In the absence of schemes unification, the situation can only be substantially changed by a process of rationalisation and coordination.
38The institutional fragmentation of old-age insurance also leads –and this is an aspect of which insured persons are scarcely aware– to major management difficulties. Today, one scheme does not have all information on their contributors’ and pensioners’ entitlements under other schemes. The schemes thus need to make financial projections of anticipated pensions and receipts without having all the information that determines the retiring behaviour and the amount of pensions. As an example civil and local government service pension schemes are not informed of the amount of state employees’ bonuses, and therefore, they cannot estimate the actual income replacement rate that they are providing for; in general, the schemes are not informed of the length of time for which their contributors have been members of other pension schemes, nor of the number of their children, or the existence or absence of spouses. This information nonetheless exists, but in the absence of coordination and pooling of information, the management of pension schemes is made particularly difficult. By 2017-2021, this Kafkaesque situation could be greatly improved through the establishment of the Répertoire de gestion des carrières unique (RGCU, “Single Career Management Register”), which aims at setting up an information system for all the schemes, making it possible to overcome the current institutional constraints.
The dependence of retirement pension expenditure on economic growth
39Today, a well-identified problem within the social protection management is the high level of dependence of pension expenditure on long-term GDP growth rates. This structural fact has been highlighted by the research.  Graph 3 sets out the impact of macroeconomic scenarios on the balance of pension schemes by 2060 according to the Conseil d’orientation des retraites (COR). Growth scenarios in productivity (and therefore in wages) play a much more decisive role than those concerning unemployment. In the optimal scenario (2% growth), the schemes will show surpluses of 1% of GDP by 2040, whereas in the low case scenario (1% growth), deficits would reach 1% of GDP in 2035. The median scenario, with 1.5% growth, yields financial balance of the schemes by 2030.
The balance of pension schemes according to growth scenarios, % of GDP
The balance of pension schemes according to growth scenarios, % of GDPField: French legally compulsory pension systems as a whole, including the old age solidarity fund (FSV) and the solidarity allowance service for elderly people (SASPA), excluding the additional public sector pension (RAFP) and the occupational hardship account (compte pénibilité).
40The mechanisms involved may be summarised as follows.  Since 1987, calculation of the reference wage used to calculate pensions of the general retirement scheme has not been revalued in accordance with wage growth but in accordance with inflation. This change, which was initially temporary, was made permanent by the reform of 2003, and led to a reduction in effective income replacement rates (but not in reference wage replacement rates, which have remained unchanged).  This index-linking mechanism has very powerful effects: in the course of time, and with growth, actual wages increase more quickly than wages recognised in the accounts (for which the increase is smaller), thus leading to a reduction in the reference wage compared to the last wage received, and this effect increases in proportion to the scale of growth. A similar mechanism is introduced into complementary schemes by indexing purchase values and entitlement values to price inflation.
41The proportion of pensions expenditure within GDP varies according to long-term growth: the quicker GDP progresses, the more the proportion of retirement pensions within GDP falls; conversely, if growth staggers for a long period, the proportion of GDP devoted to retirement pensions increases, while budgetary margins of manœuvre are simultaneously reduced.
42This method of index-linking has negative results in several respects.
43First, it gives rise to a major risk for public finances by making financial balance dependent upon the rhythm of macroeconomic growth –an uncertain variable over which public managers only have very limited control.  This condemns the schemes’ parameters to successive revisions, according to changes in the rate of economic growth.
44Furthermore, the index-linking mechanism renders the actual pattern of development of income replacement rates opaque for contributors: depending on the macroeconomic scenario, actual rates of income replacement can vary considerably (a divergence of more than 10 points for 1 negative or positive point of growth). Beyond the absence of transparency regarding the resulting implicit income replacement rate choice, there is good reason to fear that individuals may react in an inappropriate manner, by saving too much or not enough for their retirement. Uncertainty concerning actual income replacement rates reduces the value of the pension system for insured persons and therefore logically curbs their willingness to pay.
45Finally, this mechanism leads to uncontrolled redistributions between insured persons. Indeed, end-of-career wages, which are better priced, count more than older wages, a phenomenon which leads to redistributions according to the type of careers. The discrepancies between public and private sectors are also affected since public sector employees have the benefit of calculation on the basis of their last salary and are not subject to the same vagaries as private sector employees.
46This structural failing can be corrected in several different ways. Blanchet, Bozio and Rabaté (2016), op. cit. put forward three scenarios of reform in order to reduce dependence upon growth: notional accounts, the general use of a point system or a far-reaching parametrical reform. The common point of these three solutions is that of providing an adjustment mechanism that is directly dependent upon demographic change (increase of life expectancy, or a demographic ratio), without making the financial adjustment involve under-index-linking of the wages recognised in the accounts. Even within the framework of a parametric reform, the correction of this structural failing of the French retirement pensions system requires a far-reaching reform with regard to basic and supplementary schemes alike.
What reform options?
47We propose an ambitious reform of the social protection architecture, which may be envisaged in the medium to long-term. In order to take the first step in this direction, several measures may be adopted to render expenditure management more effective and transparent.
An ambitious medium to long-term reform
48An ambitious reform of French social protection would aim to resolve institutional inconsistencies without changing the current principles of our social protection. This would involve replacing the current organisation based on the institutional strata (State, basic and complementary schemes) found within each of the areas of social insurance, by a unified governance of each risk. This architecture revision would comprise three parts.
49For contributory pension rights, the reform would consist of unifying the governance of old-age insurance expenditure by the establishment of a single manager and contact, federating the existing schemes. It is important to understand that this federation does not mean the merging of schemes or the standardisation of pension rights, but unified governance, concerned with the interests of the affiliated members as a whole and providing them with a single administrative interface. The federated scheme could offer progressive harmonisation of the rules defining accumulation of pension rights, while maintaining contribution levels, and therefore differentiated entitlements. It would be a decisive step towards the convergence of schemes compared with the current trend.
50Within the framework of this federated scheme, coordination efforts could be based on the harmonisation of the various different schemes’ Information Systems, currently in progress, and its channelling into the Single Career Management Register (RGCU), in order to establish pooling of information from all schemes, accessibility of information to insured persons regarding their entitlement as a whole within the old-age insurance branch and coordinated management of pension schemes. The operational support teams (statistical, projection, etc.) for the different schemes as a whole could work together within the federal organisation and provide assistance to all of the different schemes in an effective manner.
51Recommendation 1. Unify the governance of compulsory pension schemes. Progressively harmonize the rules determining the accrual of pension rights, while, if necessary, maintaining different parameters depending on the different schemes.
52This unified governance can be envisaged in various manners. A first option would be to give Parliament a role in the management of social insurance schemes, determining the schemes’ principal annual objectives (e.g. expenditure, balance and income replacement rates), in consideration of their compulsory nature, while leaving responsibility for the management of the schemes to the social partners. Conversely, a second option would be to extend the social partners’ role to contributory social insurance as a whole (unemployment, retirement pensions, etc.), while applying the same management rules as those currently in force for complementary schemes: autonomous management, limited borrowing capacity, etc.
53As far as healthcare is concerned, the reform would concern the scope of compulsory health insurance and complementary insurance. Taking up the conclusions of Notes du CAE no 8 and 12, we propose that Parliamentary debates with regard to national investment in health should be concerned with the contents of the basket of social healthcare items, including essential healthcare which needs to be accessible to all, without any financial barriers. Access to this healthcare needs to be based upon solidarity mechanisms and should therefore be free of age-based price rates and risk selection. Unified financing of healthcare should be organised for this basket by means of decentralised public organisation or in the form of regulated competition between insurance funds.  Within this framework, the control of public healthcare expenditure would make it necessary to strike balances with regard to the content of the basket of social healthcare items, and not by means of partially bringing reimbursements to an end. Within this approach, supplementary insurance bodies’ activities would be refocused upon insurance cover for treatment outside of the basket, namely the additional insurance business.
54Recommendation 2. Unify health insurance cover (currently the social security system and complementary insurance bodies) for treatment coming within the basket of social healthcare and refocus optional insurance on cover for treatment outside of this basket.
55Finally, we propose a review of the current boundary between the respective areas of State jurisdiction and the Social Security System. A fundamental distinction is to be made between contributory benefits and benefits delivered to citizens according to their needs. The former involve financing by means of social contributions and are a part of the Social Security system, whereas the latter should be financed by taxation and fall under State authority. A review of the boundary between the State and the Social Security system is not only justified by the objective of consistency between benefits and financing, but also by the efficiency gains that can be expected from the transparency of compulsory levies connected with the opening of additional rights, thus reinforcing willingness to pay compulsory levies.
56This reform would thus lead to the inclusion of the family branch and public healthcare insurance expenditure in the State Budget, financed by taxation.  In compensation, current Social Security contributions for family and health would be entirely transferred to the old-age branch, as in the scenarios studied by the HCFi-PS. Non-allocation of income from taxation to the healthcare branch would not change the nature of the budgetary debate with regard to healthcare expenditure, organised around the ONDAM, and therefore around expenditure. When the benefits received by each individual are without any direct relation to their contributions, there is no fundamental difference between the budgetary debate concerning health and that concerning other major items of expenditure such as education and security. This financing reform would not lead to “State control” of the healthcare system, but would rather put an end to the complex and convoluted channels of revenue earmarking.
57As far as risks giving rise to contributory benefits are concerned (old age, unemployment, industrial accidents, daily allowances of sickness benefit, etc.), the current principle of operation in the form of social insurance is not to be called into question.  The reform envisaged would make the contributory character of social contributions more transparent, while avoiding the combination of different modes of financing and more clearly distinguishing these levies from ordinary taxation.
58Recommendation 3. Reorganise the architecture of social protection with a non-contributory section (family, health insurance, the fight against poverty, etc.) incorporated into the State budget and financed by taxation, and a contributory section (retirement pensions, unemployment insurance and daily allowances of sickness benefit) financed by social contributions.
59This proposal is not entirely new and it needs to be acknowledged that the structural changes of the financing of social expenditure made over the years have constituted steps in this direction.  However, this proposal remains pertinent insofar as movement is progressive and remains incomplete, whereas clearer governance would be beneficial in a context of severe budgetary constraints.
What short-term options?
60Although changing the architecture of social protection appears to be a major issue in the medium to long-term, more short-term measures can be considered in order to improve expenditure management.
Reducing the growth-dependence of pension system balance
61Without engaging in short-term structural reform, it is important to reduce the growth-dependence of long-term balance in the French pensions system by modifying the formula for the calculation of pensions in two aspects: Re-establish revaluation of the wages recognised in the accounts on the basis of average wage growth and adjust income replacement rates in accordance with demographic change, life expectancy or demographic ratio. In the case of complementary schemes, similar management can be ensured by indexing purchase values and entitlement values to wages, and also by applying a demographic coefficient to the adjustment of the income replacement rate.
62This reform does not lead to extra costs to the public purse, or to any change in average income replacement rates. It will lead to an alignment of declared and actual income replacement rates. Admittedly, the current lack of transparency is a convenient mean of reducing income replacement rates without giving rise to too much opposition. However, we are convinced that honesty in the debates is the best mean of ensuring political support of the trade-offs that need to be made and its control of the redistributive results. At the economic level, lack of transparency concerning actual income replacement rates reduces the intelligibility of reforms, which does not encourage behaviour adaptation. Moreover, uncertainty regarding long-term growth constitutes a risk for public finances, should growth prove to be lower than anticipated.
63Recommendation 4. Reduce the growth-dependence of the long-term balance of the pensions system by returning to wage indexation of salaries constituting the reference wage for pensions and by adjusting the income replacement rate with a demographic coefficient.
64Would it also be appropriate to index-link the settlement of pensions on the basis of wages rather than prices? This would make it possible to guarantee the purchasing power of the most elderly persons, whose consumption is to a large extent focused on personal services, and further reduce the growth-dependence of the balance of the system. However, at constant levels of expenditure, this more dynamic progress of pensions would need to be offset by a reduction in the income replacement rates paid out, which would render its acceptability problematic. Furthermore, this choice of index-linking for the settlement of pensions would reinforce redistribution from retired persons with few years to live to those who live longer.
For healthcare, put a ceiling on remaining payable costs for recognised expenditure
65Currently, complementary insurance bodies’ activity combines complementary cover for recognised expenditure and additional cover for items not falling within this basket. Regulation of the sector does not distinguish between these two areas. However, solidarity mechanisms need to be applied as to the complementary part of cover, while there is no reason to do so for the additional part. In order to protect persons covered by social insurance, the capping of remaining payable costs would constitute an effective mechanism: individuals whose combined annual remaining payable costs exceed a given amount would be eligible for 100% public insurance cover within the field of recognised expenditure. In this case, public expenditure would provide protective cover for all against catastrophic expenditure. This protection would be effective insofar as regulation of the sector would guarantee access to provision of treatment at statutory rates valid throughout French territory. 
66Recommendation 5. Establish public cover of 100% of remaining payable costs for recognised expenditure beyond a certain limit. Guarantee access to provision of treatment at statutory rates valid throughout French territory.
Promote the democratic decision-making process
67In order to promote Parliamentary debate on the basis of risk coverage, and without awaiting far-reaching institutional reform, we propose the inclusion of budgetary documents in a breakdown of public expenditure as a whole according to the function or risk covered.  This would make it possible to clearly place the progression of pension, health, and unemployment expenditure within public debate, alongside changes in public expenditure on the part of the State and local authorities.
68Recommendation 6. Present an annexe every year, at the time of budgetary debates, setting out a breakdown of public expenditure according to function and the social risk covered.
69The inconsistencies of our social protection system have become major obstacles to effective and democratic management of social protection expenditure. Although an ambitious reform will take time, it is inevitable if we wish to maintain the high levels of cover that we know.
Paris School of Economics (PSE), École des hautes études en sciences sociales (EHESS), Institut des politiques publiques (IPP), Member of the CAE.
PSL, University Paris Dauphine (LEDa), Member of the CAE.
Data for 2013 from Comptes de la protection sociale 2015.
European System of integrated Social Protection Statistics.
The field includes compulsory schemes, group insurance and benefits paid by mutual insurance companies governed by the Mutual Insurance Code (Code de la mutualité), as well as supplementary health insurance provided by insurance companies, whose pricing does not a priori come within the scope of solidarity mechanisms.
Eurostat ESSPROS data and the OECD SOCX database.
To our knowledge no country has a social protection architecture, in particular concerning pensions, as complex as France at the level public administrations. Abroad most of the complexity results from competing optional schemes.
Complementary group schemes, which have been compulsory in private companies since 1st January 2016, do not fall within the field of general government since the contracts are drawn up based on a choice of the service provider, a certain amount of latitude regarding the choice of coverage and financial autonomy of the service provider in relation to general government. Dormont B., P-Y. Geoffard and J. Tirole (2014): “Rebuilding the Health Insurance System”, Note du CAE, no 12, April.
See the Report of the Haut conseil du financement de la protection sociale (HCFi-PS) (2013): Rapport d’étape sur la clarification et la diversification des ressources des régimes de la protection sociale, June, for a discussion of the principles of contributiveness, universality and redistribution.
For a review of the literature see Saez E., J. Slemrod and S.H. Giertz (2012): “The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review”, Journal of Economic Literature, vol. 50, no 1, pp. 3-50.
Wilkinson M. (1994): “Paying for Public Spending: Is There a Role for Earmarked Taxes?”, Fiscal Studies, vol. 15, no 4, pp. 119 35.
It should be noted that the twofold legal nature of the CSG in domestic law and European Union law reveals this ambiguity: the CSG was defined as a general tax in domestic law because of its intended purpose, whereas it is considered equivalent to a social security contribution in European Union law, in view of its allocation to the social security system.
HCFi-PS (2013) op. cit. and HCFi-PS (2015): Rapport sur la lisibilité des prélèvements et l’architecture financière des régimes sociaux, July.
(0,05 + 0,17) / (1 – 0,09) = 0,24.
Moreover, for indisputable ethical reasons, the State cannot refuse the provision of healthcare for uninsured persons. It is therefore logical to oblige individuals to contribute to the common fund ex ante.
Consumption of medical care and materials encompasses “all consumption aimed at the treatment of a disturbance in the patient’s state of health”, cf. Beffy M., R. Roussel, J. Solard and M. Mikou (2015): “Les dépenses de santé en 2014”, Études et Statistiques, DREES, no 935, September.
See the examination in Note du CAE no 8: Askenazy P., B. Dormont, P-Y. Geoffard and V. Paris (2013): “Towards a More Efficient Health System”, Note du CAE, no 8, July.
In other words, if nothing is changed in the coverage rate of different treatments by the social security system, due to a structural effect connected with the increase in the proportion of insured persons with long-term illnesses, there is a progressive increase in the average coverage rate. Various different contributions to patients’ expenses were introduced before 2012 in order to counteract the effects of this increase.
There is a recent and very positive example of expenditure management by using the healthcare basket in the field of medicines, in which price reduction in the development of generics has enabled wide and rapid distribution of a major innovation in the treatment of hepatitis C (sofosbuvir), see www.ameli.fr/fileadmin/user_upload/documents/cnamts_rapport_charges_ produits_2016.pdf
Célant N., P. Dourgnon, S. Guillaume, A. Pierre, T. Rochereau and C. Sermet (2014): “L’enquête santé et protection sociale (ESPS) 2012. Premiers”, Questions d’Économie de la Santé, IRDES, no 198, May.
Amounts not covered by the Social Security system among items of expenditure giving rise to reimbursement entitlements: costs not reimbursed by the social-security system, fixed-rate contributions, “excesses”, fixed hospital rates and excesses beyond valid price rates.
Cf. DREES, Échantillon interrégimes retraites (EIR) 2012, calculations by the authors.
World Bank (1994): Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth, Washington DC.
See in particular Blanchet D. (2013): “Retraites : vers l’équilibre en longue période ?”, Note IPP, no 3 or Marino A. (2014): “Vingt ans de réformes des retraites : quelle contribution des règles d’indexation ?”, INSEE Analyses, no 17.
See, in particular, Blanchet D., A. Bozio and S. Rabaté (2016): “Quelles options pour réduire la dépendance à la croissance du système de retraite français ?”, Revue Économique, forthcoming.
The actual rate of income replacement is the ratio between the average pension received in the year of settlement and the average income received in the final year of activity. Today, only Belgium, Spain and France revalue the wages recognised in the account according to inflation.
Index-linking to prices admittedly limits the growth of pension expenditure, but its actual impact upon public finances remains uncontrolled, and dependent on economic growth.
It should be emphasised that two forms of organisation are compatible with financing using taxation: in any case, revenue allocation is effected in the form of a head tax for each person covered, equal to their expected healthcare expenditure taking their observable characteristics into account for the calculation. Furthermore, decentralised public financing refers to the possibility of diversification of payment methods and to the possibility of content of official approval granted to doctors by the Department of Health for the reimbursement of medical expenses according to the different territories, see Note CAE no. 8 op. cit. The scenario of different rates of contribution and reimbursement is excluded.
Means-tested benefits also appear in the State budget with the aim of fighting against poverty and social exclusion, which have always been financed by taxation.
So called non-contributory mechanisms within the old-age branch and with regard to unemployment cover should remain under State governance, with financing by means of taxation.
The most recent example is the provision in the 2016 Budget bill (PLF) for “inclusion in the Budget” of the majority of housing benefits.
This needs to involve a policy of approval of healthcare providers by the Department of Health for reimbursement purposes and establishment of contracts, which is essential in order to ensure that the provision of healthcare is managed in accordance with needs, see Notes du CAE nos. 8 and 12, op. cit.
Within the framework of the Institutional Act concerning Budgets (LOLF: Loi organique relative aux lois de finances), State expenditure is presented according to missions and programs in order to provide an overall interpretation of public policies: however, this only includes the State budget rather than the public administrations as a whole.