Déchiffrer 2027

FREN

method

How we compare and cost the programmes

We record what each candidate says, translate it into parameters relative to current law, then cost it with the same cost schedule for all. Every step is published and open to challenge.

What these figures are worth

The four stages

  1. Collection: what does the candidate say? The exact quote, its source and date, and who is speaking: the written programme, a statement by the candidate, an associate, or a document of their party. One position per lever: where there are duplicates, we keep the strongest source (programme, then statement, then associate, then party), then the most recent.
  2. Encoding: what does it change? The direction of change relative to current law (increase, decrease, unchanged, abolition, creation), then, where possible, the parameter affected, its current value taken from the cost schedule and the proposed value. When the text is vague and we have to choose a reading, the page says so.
  3. Costing: how much does it cost or raise? An annual order of magnitude, in a full year, relative to the continuation of current policies, with a low, central and high range, in euros of the year of the source of each coefficient (shown on each page). It comes from a schedule of unit costs published below, the same for everyone, and it is compared with the figure announced by the candidate when they give one.
  4. Effects: who gains, who loses? The effect on the disposable income of eight typical households, calculated with OpenFisca-France for the positions whose encoding gives a value that the engine can apply: see My household. The effect by standard of living is not yet online.

The costing rules

What an amount counts and does not count, when a position remains not costed, what assumptions the editors may make, how values outside the range of the sources and overlaps are handled, and on what condition a total is published.

The same rules apply to all candidates. The costing engine alone applies the calculation, the extrapolation, the overlaps, the comparison with the announced amounts and the “assumption” mark on coefficients calculated by the editors. Four marks are set position by position by the editors, with their reason published on the page: the central value moved to one end of the range (acknowledged bias), the assumption on how to read a text, partial costing, and the statement that a source costs one position separately from another (measured overlap).

  • Static. We count the direct effect on the public accounts, with no behavioural effect and no feedback on growth. When these effects matter a great deal (pensions, employment), the cost schedule cites them separately, never in the central figure. What this choice assumes: models and feedback effects.
  • Relative to current law. A measure already voted or scheduled (the abolition of the CVAE, a business value-added contribution, in 2030, for example) costs nothing more; keeping it beyond its end date raises revenue. For defence, the reference is the latest level set by the military programming act in force (2.5% of GDP in 2030), maintained in 2032: a candidate who adopts it is at the status quo.
  • How a position is linked to the cost schedule. Either a linear parameter (“each additional month of benefit duration costs…”) with the value given by the position, or a measure costed as such by the source (an option), when the position matches it. Otherwise, the position is not costed, and the page says why. A position costed at €0 (a value equal to current law, a measure with no static cost) counts as a status quo.
  • The most recent source prevails. When the same candidate (or their party’s programme, taken as their source) has said something else on the same subject in a more recent source, that source counts: a programme that leaves a question “not settled” makes the older proposal of an op-ed not costed, and a commitment not to raise any tax makes a tax increase proposed earlier not costed. The older source remains cited, with the reason.
  • The permitted assumptions, and only those. The editors may convert an announced value into the unit of the parameter (a gross SMIC into net, a five-year envelope into an annual amount, a ten-year target brought back to its full year), or adopt one reading of a vague text (the note says which one and, if there is another, cites it). The position is then marked “editors’ assumption”. An assumption never adds up two costings from different sources, horizons or reference laws.
  • “At least”, “up to”, “from X to Y”. The position is costed at the most cautious value cited (the floor of “at least”, the lower bound of a span) and marked “partial costing”. A value cited as such (retirement at 60) is costed at that value, never replaced by another.
  • Measured range and extrapolation (*). Each parameter has a range: the span that its source measured (a retirement age between 62 and 64, for example). A proposed value beyond it is still costed, by extending the coefficient in a straight line: the line carries an asterisk, “outside the range measured by the source: linear extrapolation”, on the page, in the costing summary and in the downloadable data, and the costing page of each candidate lists their extrapolated lines. It counts in the totals and in the coverage. An option, costed as such by its source, is not extrapolated. When the source costs the increase and the decrease of a parameter at different values, each direction has its own coefficient: for the contribution period, the Cour des comptes (Court of Audit) costs one more quarter at €1.3bn in 2035 and one quarter less at about €1bn; the cost schedule uses €0.75bn per quarter for the decrease, a value brought back to 2032, and the 2035 figure for the increase, which slightly overstates the gain in 2032. A position that changes several parameters of the same lever (reduced VAT rates merged into one) adds them up.
  • Coefficients calculated by the editors. When a primary source publishes the yield of a levy (corporate income tax, excise duties, CSG, digital services tax…) without giving the cost of one point, the editors calculate it (yield ÷ rate); the cost schedule shows it as a “coefficient calculated by the editors from” its source, and any position costed with such a coefficient is marked “editors’ assumption” by the engine itself. A primary source that costs the measure directly replaces this calculation (the yield of one point of each VAT rate published by the DG Trésor, the increase in the financial transactions tax costed by the Senate).
  • Acknowledged bias in the costing. When the note on a position acknowledges a bias in its costing, the central value goes to the end of the coefficient’s range that corrects it: down to the smallest amount for an overestimate, up to the largest amount for an underestimate (an extrapolation that does not count the whole cost, for example). The range stays the same; the page, the costing page and the downloadable data say so, with the reason. A candidate’s “at least” is not a bias in the costing: it remains costed at the floor. The rule applies to all candidates.
  • Earmarked revenue. When the same sentence earmarks a costed revenue for a named expenditure or tax cut (the proceeds of a tax “earmarked” for a fund), the counterpart is counted with it, at the same amount as the revenue when the text gives none: the revenue goes into new revenue, the expenditure into new spending, and the net effect on the balance is roughly nil.
  • Financing that cannot be costed. Every costable measure is counted. When the candidate finances it with revenue that cannot be costed (neither rate nor schedule) or with overall savings that are not detailed, this financing is shown as an announced amount, with the reason, without being added: the measure therefore appears without its counterpart, and the page says so. The same rule applies to all candidates (the rule of 18 October 2026 that also removed the measure, when its revenue could not be costed, is cancelled: it did not apply to financing through overall savings).
  • Amounts taken from the candidate. When a parameter is only an envelope (“€1bn announced is worth €1bn”: appropriations, aid, investment), the amount is the candidate’s, “taken as is”, and no gap is calculated against their announcement.
  • Zucman tax. The 2% minimum tax on assets above €100m is costed and added, for all the candidates who propose it, at the upper end of the estimates: €15bn to €25bn (€20bn central), the static estimate of its proponents (Gabriel Zucman, reports of the National Assembly No. 930 and of the Senate). The administration has not simulated it; the estimates that take account of exile, tax optimisation or the deduction of corporate income tax are much lower (€2bn to €5bn). The amount counted is therefore an upper bound, and the line says so.
  • Net in the centre, gross as the upper bound. The central value is the net yield for general government as a whole when a source read gives it or allows it to be deduced (VAT, CSG, pension indexation); the gross yield is used as the upper bound. For public sector staff, the centre counts the cost excluding contributions paid to other public administrations (State pensions, internal employer contributions) and the upper bound the full employer cost; the same convention applies to flows between public bodies. Thus, one public sector pay index point costs about €1.5bn in the centre, excluding contributions paid to other administrations (net convention), and about €2.1bn to €2.3bn in gross employer cost, as the upper bound. A range never mixes two scopes, nor a static costing with a costing that includes behavioural effects.
  • One measure, one lever; overlaps. The same measure is costed only once (the 10% allowance for pensioners under the lever that carries it, not also among tax expenditures). The cost schedule declares the costings that overlap, from one lever to another (an envelope and one of its items) or between two options of the same lever. For the same candidate, if a source costs one of the positions separately from the other, both count. Otherwise, the cost schedule estimates, for each, the share of its amount that falls within the common base (the public sector pay bill around the minimum index, for the SMIC and the index point), with justification and source: both amounts count, and only the common part, the smaller of the two products “share × amount”, bound by bound, is removed from the smaller of the two positions, which shows this amount as “not added” and is marked “partial costing”. Each bound keeps its place: when the common part varies from one bound to another to the point where removing it from the smaller position would invert its bounds (its central value falling below its lower bound), it is removed from the larger one, and the sum of the two, bound by bound, stays the same. The shares may depend on the direction of the two amounts: the common share of two increases (the legal retirement age and the contribution period raised together) is not that of two decreases. When the base of the smaller position is entirely included in the other, it is shown without being added; when the common base is nil (or when no source says that an overall staffing target is counted net of the posts created in a sector), the two are added together. In the absence of an estimate of the common base, between two costs or two gains, only the higher of the two amounts is counted, the other is shown as “not added” and the counted line is marked “partial costing”; a cost and a gain are always added together.
  • Assumptions, partial costings and announced amounts. When the editors had to choose a reading, or when part of the position is not costed, the page and the costing summary say so. All the amounts announced by a candidate are cited on their costing page, with the meaning they give them (spending, revenue cut, revenue, savings), never deducted from the lever, and without being added to the totals. The gap with our calculation is calculated only if both cover the same scope: not for an assumption, a partial costing, an amount taken as is, or an announcement that covers another year or another scope. It is calculated only if the announcement, read at its own precision (€25bn announced means €24.5bn to €25.5bn), falls outside the range of our calculation; an announcement that touches it is said to be “within the range”, with no gap.
  • Horizon and euros. Each amount is an annual order of magnitude in a full year, in euros of the year of its source; the page of each position states that year. The sources date from 1986 to 2026: a recalibration of all the coefficients on a single price and GDP framework is planned.
  • Costing summary. For each programme: new spending, revenue cuts, new revenue, savings; each amount of a position goes into the total of its nature (a revenue and the expenditure it finances, in the same position, each count in their own). Loans and asset purchases, which weigh on the debt and not on the deficit (the financing of the EPR2 reactors), are counted separately. These totals are published for all candidates; under each, the sum of the lower bounds and that of the upper bounds of the lines (ranges are added bound by bound, which gives an order of magnitude, not a confidence interval). Their sum, the balance, is published only if at least 60% of the essential costable levers carry a position of the candidate whose cost is known, a position costed only in part counting for half: below that, it would add up scopes that bear no relation from one candidate to another. The grid never shows a total, only the costed share. A more costly total does not in itself make a worse programme: the amounts are static, with no feedback effect, and each programme rests on an economic model that the cost schedule does not decide between.
  • Coverage. An essential lever counts as covered when the candidate’s position has a known cost: costed (even partly, by extrapolation, or not added because another position covers it) or status quo. The costing summary also states what share of the positions recorded could be costed (status quo positions, at zero cost, are counted separately) and how many remain not costed on the revenue side and on the spending side. A programme covered at 15% is never presented as complete.

Models and feedback effects

Our amounts are static: what published models say about feedback effects, why they diverge, and why the site publishes no cost “after feedback effects” per candidate.

The site’s amounts are static: they count what a measure costs or raises for the public accounts, without its effect on economic activity. This choice is not neutral: it amounts to assuming that this effect is nil. This section sets out what published economic models make of it, and why they do not agree.

Multiplier and feedback effect

A fiscal multiplier measures the effect on GDP of one euro of a measure: a multiplier of 0.8 means that one more euro of spending, or one euro less of tax, increases output by 80 cents, all other things being equal. This additional activity brings in taxes and contributions: this is the feedback effect, of the order of the effect on GDP multiplied by the rate of compulsory levies (43.6% of GDP in 20251).

The effect works in both directions. New spending or a tax cut supports activity and thus recovers part of its cost; savings or a tax increase slow activity and lose part of their yield. A stimulus programme therefore costs less than its static amount, and a consolidation programme raises less than its own, in proportions that depend on the multipliers.

Published multipliers, by category of measure

Fiscal multipliers by category of measure and by horizon, central value and range of the sources
Category of measure1 to 2 years5 years10 years
Public investment
11 sources, with the excerpt read
1.3 (0.8 to 1.4)0.9 (0.85 to 0.95)1.3 (0.5 to 2.8)
Housing investment aid (renovation, social housing)
2 sources, with the excerpt read
1.3 (0.8 to 1.4)0.9 (0.85 to 1)0.8 (0.55 to 1)
Research and innovation support (public research, research tax credit, France 2030)
5 sources, with the excerpt read
—no value of its own—no value of its own0.7 (0 to 1.5)50 to 100 years: 3.3 (2.3 to 4.4)
Operating expenditure (public sector wages, purchases, services)
9 sources, with the excerpt read
0.9 (0.7 to 1.05)0.9 (0.85 to 0.9)0.6 (0.55 to 0.8)
Transfers to low-income households (minimum social benefits, in-work benefit, APL, unemployment benefit)
6 sources, with the excerpt read
0.8 (0.6 to 1)0.8 (0.6 to 1)0.6 (0.55 to 0.8)
Other transfers to households (pensions, family allowances, other benefits)
5 sources, with the excerpt read
0.6 (0.4 to 0.85)0.6 (0.4 to 0.85)0.6 (0.55 to 0.8)
Tax cut targeted at low-income households
6 sources, with the excerpt read
0.8 (0.6 to 1)0.8 (0.6 to 1)0.9 (0.8 to 1.35)
Broad-based household taxes (VAT, CSG, excise duties, income tax, property tax), decrease or increase
8 sources, with the excerpt read
0.5 (0.3 to 0.85)0.9 (0.8 to 1.1)0.9 (0.8 to 1.35)
Taxes on well-off households (capital, wealth, inheritance), decrease or increase
6 sources, with the excerpt read
  • OFCE, Policy brief n° 149, 15 octobre 2025, p. 5, 2025-10-15 : « Comme l’effort budgétaire s’explique essentiellement par les hausses de fiscalité en 2025, ciblées sur les grands groupes et les hauts revenus, le multiplicateur budgétaire associé à la politique budgétaire serait relativement faible à court terme (évalué à un peu moins de 0,5 à un an). »
  • Insee, Note de conjoncture de juin 2019, dossier « Quel lien entre pouvoir d’achat et consommation des ménages en France aujourd’hui ? », p. 27 et 35, 2019-06 : « À l’inverse, les ménages les plus aisés ont épargné près de 40 % de leur revenu (graphique 6). […] la PMC des ménages les plus aisés est la plus faible donc l’évolution de leur revenu a une incidence beaucoup plus modérée sur la consommation. »
  • Institut Avant-garde (note d'A.-L. Delatte, CNRS), Redresser les finances publiques sans sacrifier l’avenir : une simulation macro-budgétaire, p. 15, 2025-10 : « les multiplicateurs de l’impôt sur les sociétés (IS) et de la taxation du patrimoine ne sont pas estimés dans Mésange. Le premier provient d’une estimation de l’OFCE13, tandis que la faible valeur retenue pour le second (0,2) s’explique non seulement par une faible propension marginale à consommer (PMC), mais surtout par un effet très limité sur l’investissement […] Un exercice de sensibilité est présenté en annexe, avec un multiplicateur alternatif de 0,5. »
  • France Stratégie, Comité d’évaluation des réformes de la fiscalité du capital, troisième rapport, p. 16, 40 et 115, 2021-10 : « le comité considère que la majorité de cette augmentation a bien été causée par l’instauration du PFU. […] la plupart des travaux empiriques convergent pour conclure à l’existence d’un effet négatif significatif sur l’investissement du niveau de l’impôt sur les sociétés, mais à l’absence d’un effet tangible du niveau d’imposition qui s’exerce sur les revenus des actionnaires (dividendes en particulier). […] La littérature sur la taxation du patrimoine des ménages semble indiquer que la taxation du patrimoine a un impact durable sur son accumulation par les ménages les plus aisés. […] De même qu’en 2013, aucun effet significatif sur l’investissement de ces entreprises n’était détecté. »
  • Institut Avant-garde (note d'A.-L. Delatte, CNRS), Redresser les finances publiques sans sacrifier l’avenir : une simulation macro-budgétaire, tableau 7, p. 15-16 (année 6), 2025-10 : « Les années 4 et 6, non estimées par le modèle Mésange, sont obtenues par interpolation linéaire : l’année 4 est la moyenne des années 3 et 5 ; l’année 6 prolonge l’écart observé entre les années 4 et 5. […] Tableau 7 – Multiplicateurs associés aux mesures Mesure Année 1 Année 2 Année 3 Année 4 Année 5 Année 6 […] IS -0,2 -0,3 -0,4 -0,4 -0,4 -0,4 Patrimoine -0,2 -0,2 -0,2 -0,2 -0,2 -0,2 »
  • France Stratégie, Comité d’évaluation des réformes de la fiscalité du capital, troisième rapport, p. 40, 2021-10 : « la plupart des travaux empiriques convergent pour conclure à l’existence d’un effet négatif significatif sur l’investissement du niveau de l’impôt sur les sociétés, mais à l’absence d’un effet tangible du niveau d’imposition qui s’exerce sur les revenus des actionnaires (dividendes en particulier). […] La littérature sur la taxation du patrimoine des ménages semble indiquer que la taxation du patrimoine a un impact durable sur son accumulation par les ménages les plus aisés. »
0.3 (0.2 to 0.45)0.3 (0.2 to 0.45)0.3 (0.2 to 0.45)
Levies on businesses (employer contributions, IS, production taxes) and aid to businesses, decrease or increase
11 sources, with the excerpt read
0.3 (0.2 to 0.55)0.8 (0.4 to 0.95)1 (0.4 to 1.3)

Effect on the level of GDP, in euros, of one euro of a permanent measure, in the year shown: central value and, in brackets, the smallest and the largest of the values given by the sources read. An increase and a decrease of the same levy have the same multiplier, with the opposite sign; savings take the multiplier of the spending of the same nature. When no source gives a horizon, the value of the previous horizon is carried over. Research has a value of its own only from the sixth year: before that, each measure keeps its category (operating expenditure for public research, businesses for the research tax credit). The sources for each row, with the excerpt read, are under its heading; survey closed on 15 October 2026.

Why they diverge

A multiplier is not an observed constant: it is the result of a model, which depends on what that model assumes. Four assumptions weigh more than the others.

  • The long term is set by supply. Mésange, the model of the DG Trésor and Insee, has « une dynamique keynésienne à court terme et un équilibre de long terme davantage déterminé par des facteurs d’offre » [“Keynesian dynamics in the short term and a long-term equilibrium determined more by supply factors”]2. The effect of spending fades in it: a public spending shock made up of 46% cash benefits raises GDP by 0.76% in the first year and by 0.17% in the long term5. The model also reasons at an unchanged real interest rate6.
  • Public capital is not productive in it. For public investment, Mésange describes « un pur choc de demande car, dans le modèle Mésange, la productivité du stock de capital public n’est pas prise en compte » [“a pure demand shock because, in the Mésange model, the productivity of the public capital stock is not taken into account”]: 1.31 in the first year, 0.22 in the long term3. The OFCE adds a supply effect, assessed at 0.5 « à terme » [“in the long run”] in a worked example7. The IMF (A. Abiad, D. Furceri, P. Topalova, 2015) estimates, for advanced economies, that one point of GDP of public investment raises output by 1.5% four years later, i.e. a multiplier of about 1.4, from 0.7% to 2.6% depending on the efficiency of the investment8.
  • Supply effects for some, not for others. In Mésange, a cut in employer contributions raises GDP by 0.20% in the first year and by 1.73% in the long term (+362,000 jobs), through the lower cost of labour4. The published variants we have read give no supply effect of their own to benefits, nor to spending on education or health.
  • Is lost activity made up? Another body of literature, on hysteresis, argues that a recession or a consolidation can leave lasting marks on employment and on productive capacity (long-term unemployment, investment not made). If so, the effect of a demand measure does not fade entirely in the long term.

Innovation and human capital. Innovation-driven growth earned Joel Mokyr, Philippe Aghion and Peter Howitt the 2025 Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel, awarded “for having explained innovation-driven economic growth”9. In QUEST III, the European Commission’s endogenous growth model, a permanent subsidy to research does not raise GDP in the first ten years (a nil or slightly negative effect), then raises it by 2.3 to 4.4 euros per euro spent after 50 to 100 years10. For education and health, none of the public sources we have read links one euro of spending to a quantified effect on activity: France Stratégie quantifies what better school results would bring in, but considers it necessary « d’identifier les réformes éducatives qui permettraient d’améliorer durablement et significativement les performances des élèves » [“to identify the educational reforms that would durably and significantly improve pupils’ performance”]11.

What the site draws from this

The cost of a programme cannot be read on its own. Depending on the model adopted, the same euro of spending or savings produces very different feedback effects, and the gap widens with the horizon: for public investment, from 0.22 in the long term in Mésange to about 1.4 four years later according to the IMF. The site therefore does not publish a cost “after feedback effects” per candidate: the result would depend on the model chosen, that is, on a choice that the site does not make, and would leave aside what no multiplier measures (public services, inequality, climate, external dependence). Each costing summary cites the economic model the candidate claims, without endorsing it, and states what the costing does not measure. See also what these figures are worth.

Sources for the text

  1. Insee, Insee Première n° 2106, Le compte des administrations publiques en 2025, 2026-05: « Le taux de prélèvements obligatoires (net des crédits d’impôt) augmente de 0,9 point et atteint 43,6 % du PIB en 2025. »
  2. DG Trésor et Insee (modèle Mésange), Document de travail n° 2017/04, Le modèle macroéconométrique Mésange : réestimation et nouveautés, résumé, p. 4, 2017-05: « Mésange se caractérise par une dynamique keynésienne à court terme et un équilibre de long terme davantage déterminé par des facteurs d’offre. »
  3. DG Trésor et Insee (modèle Mésange), Document de travail n° 2017/04, Le modèle macroéconométrique Mésange : réestimation et nouveautés, § 4.1.1, p. 102-103, 2017-05: « Hausse de l’investissement public de 1 % du PIB en volume % en écart au compte central 1 an 2 ans 3 ans 5 ans 10 ans LT PIB en volume 1,31 1,38 1,22 0,93 0,57 0,22 […] Il s’agit d’un pur choc de demande car, dans le modèle Mésange, la productivité du stock de capital public n’est pas prise en compte »
  4. DG Trésor et Insee (modèle Mésange), Document de travail n° 2017/04, Le modèle macroéconométrique Mésange : réestimation et nouveautés, § 4.1.3, p. 107-108, 2017-05: « Le gain d’activité lié à la baisse du coût du travail est présent dès la première année (+0,20 % sur le PIB et +83 000 emplois créés) et s’amplifie par la suite (+0,97 % sur le PIB à 5 ans et +316 000 emplois) et ce jusqu’au long terme (+1,73 % sur le PIB et +362 000 emplois). »
  5. DG Trésor et OFCE, Document de travail n° 2026/1, Comparaison des propriétés des modèles macroéconomiques français ThreeME et Mésange, § 2.2, p. 18-19, 2026-07: « Dans Mésange, le choc est constitué d’investissements publics (8 %), de rémunérations publiques (25 %), de dépenses de prestations sociales en nature des administrations publiques − APU (9 %), de consommations intermédiaires des APU (12 %), et de prestations sociales en espèces (46 %). […] Hausse des dépenses publiques de 1 point de PIB ThreeME-France standard Mésange t t + 1 t + 4 t + 9 t + 29 LT t t + 1 t + 4 t + 9 t + 29 LT PIB % 0,68 0,79 0,90 0,80 0,46 0,45 0,76 1,01 0,88 0,54 0,27 0,17 »
  6. DG Trésor et Insee (modèle Mésange), Document de travail n° 2017/04, Le modèle macroéconométrique Mésange : réestimation et nouveautés, p. 103, note 73, 2017-05: « À noter qu’il n’y a pas dans le modèle Mésange d’éviction par les taux d’intérêt comme dans un modèle IS/LM usuel car les variantes sont réalisées à taux d’intérêt réel inchangé. »
  7. OFCE, Policy brief n° 25, Évaluation du programme présidentiel pour le quinquennat 2017-2022, p. 9, note 3, 2017-07-12: « Par exemple, en retenant un déclassement sur 20 ans, soit un taux de déclassement de 5 %, pour une économie à 25 % de son optimum, le multiplicateur d’investissement public au sens large (y compris dépenses d’éducation) s’élève à 0,5 à terme. »
  8. FMI, Working Paper WP/15/95, The Macroeconomic Effects of Public Investment: Evidence from Advanced Economies (A. Abiad, D. Furceri, P. Topalova), p. 10 et 13, 2015-05: « Four years after an unanticipated shock to government investment spending of 1 percentage point of GDP, the level of real output is 1.5 percent higher, which corresponds to a medium-term fiscal multiplier of about 1.4. […] In countries with high efficiency of public investment, a public investment spending shock increases the level of output by about 0.8 percent in the same year and by 2.6 percent four years after the shock. But in countries with low efficiency of public investment, the output effect is about 0.2 percent in the same year and about 0.7 percent in the medium term. »
  9. Nobel Prize Outreach (nobelprize.org), The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2025, summary, 2025-10: « Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2025 was awarded "for having explained innovation-driven economic growth" with one half to Joel Mokyr "for having identified the prerequisites for sustained growth through technological progress" and the other half jointly to Philippe Aghion and Peter Howitt "for the theory of sustained growth through creative destruction" »
  10. Commission européenne (DG ECFIN, modèle QUEST III à croissance endogène), Economic Papers n° 351, Structural reforms in the EU: a simulation-based analysis using the QUEST model with endogenous growth (W. Roeger, J. Varga, J. in 't Veld), § 3.1, tableaux 3.1.a et 3.1.b, p. 21-24, 2008-12: « Table 3.1.a 0.1% of GDP tax-credit R&D subsidy to the non-liquidity constrained households Years EU 1 2 3 4 5 10 20 50 100 GDP -0.01 -0.04 -0.05 -0.06 -0.05 0.00 0.08 0.23 0.31 […] Table 3.1.b 0.1% of GDP wage subsidy to the R&D sector Years EU 1 2 3 4 5 10 20 50 100 GDP -0.02 -0.06 -0.08 -0.08 -0.08 -0.01 0.11 0.33 0.44 »
  11. France Stratégie, Note d'analyse n° 48, Investir dans l'éducation pour favoriser la croissance (juin 2016), p. 1 (copie hébergée par la Fondation pour l'enfance), 2016-06: « De telles estimations sont néanmoins à considérer avec prudence, car il reste délicat de quantifier l’impact causal de la qualité du capital humain sur la croissance. Au-delà de cette difficulté, il est également nécessaire d’identifier les réformes éducatives qui permettraient d’améliorer durablement et significativement les performances des élèves »

Which candidates, and which by default

Who appears in the grid, and the written rule that sets the candidates shown by default.

Positions are recorded for people whose candidacy is declared or who are nominated by their party, and who are credited with at least 2% in one of the scenarios of the latest voting intentions poll recorded (Elabe for BFMTV and La Tribune Dimanche, online survey from 14 to 16 October 2026, 1,701 people of whom 1,553 registered voters intending to vote; Notice déposée par Elabe à la Commission des sondages (n° 10324), 17 October 2026). A released programme is not enough to enter the grid; the other declared candidacies are mentioned on the candidates page, with their programme when one exists. The same criterion applies to all; it will be reapplied with each new poll. As of 17 October 2026, in the grid: Gabriel Attal, Nicolas Dupont-Aignan, Raphaël Glucksmann, Marine Le Pen, David Lisnard, Jean-Luc Mélenchon, Édouard Philippe, Bruno Retailleau, Fabien Roussel, François Ruffin, Marine Tondelier, Éric Zemmour.

Every candidate covered appears in the grid, whether or not they have published a programme; without a released programme, their positions are taken from their statements, and their column says so. By default, the top six in the latest poll recorded: Elabe for BFMTV and La Tribune Dimanche, 14 to 16 October 2026 (Notice déposée par Elabe à la Commission des sondages (n° 10324), 17 October 2026).

Anyone can change the candidates shown in the grid, up to six at a time. Details of statuses and programmes are in Who costs?.

Candidate pages

The same template for all candidate pages, and the sources required for each section, including justice.

Each candidate in the grid has a page, with the same template for all: recent appearances, schedule, career, transparency, justice. A section with nothing in it says “Nothing recorded to date”, never an asserted absence. A page not yet written says so.

  • Sources. Each item links to a source that has been read and dated. Offices held rely on an official source (National Assembly, Senate, European Parliament, Journal officiel, the local authority’s website); declarations link to the Haute Autorité pour la transparence de la vie publique (HATVP, High Authority for Transparency in Public Life), without copying their content; campaign accounts, to the decisions of the CNCCFP (national commission on campaign accounts) or the Journal officiel.
  • Appearances. Recent interviews, op-eds and speeches by the candidate or by associates (spokespersons, party leadership), with programme content. Quotes are copied word for word from the source; those of an associate do not formally commit the candidate, except those of the person they have tasked with carrying their programme (for example their head of economic programme), presented as the candidate’s own with the name of the speaker.
  • English version. The site is translated by the editors. Candidates’ quotes, sources and videos stay in their original French; each quote is followed by the editors’ translation, labelled “Translation by the editors”, which is never presented as the speaker’s own words. Inside editorial notes, quoted French words are kept in « » with an English rendering in square brackets. A text not yet translated appears in French with the mention “Not yet translated”.
  • Schedule. Events announced by the campaign, the party or the reference press. An event more than 60 days in the past is no longer shown.
  • Justice. Criminal proceedings in which the person is personally implicated, ongoing or whose latest decision is less than ten years old, acquittals and dismissals included. Each stage (mise en examen, referral to trial, judgment, appeal, appeal to the Cour de cassation, final decision) is dated and rests on an official decision or press release, or on at least two reference media that cite the decision. The sentence is reproduced in the court’s terms when the decision is published; otherwise, it is attributed to the media outlet that reports it. An appeal decision delivered on the sole appeal of a civil party is presented as such: it concerns civil interests, not guilt. Private individuals who are implicated or who are complainants are not named. The section includes no simple complaint, no suspicion, no investigative article without a decision, and no characterisation. Any proceedings that are not final carry a reminder of the presumption of innocence. The section is not included in the downloadable files: it is read on the page, up to date.
  • Checking. Each page shows the date it was checked. An error, an omission or a more recent decision can be reported through the “Challenge or complete” link at the bottom of the page; the correction is published.

The 2017-2027 record

Where the record’s indicators come from, their starting point, their comparisons, and why the page attributes no change to a policy without an official source.

The Record page brings together the official indicators of Emmanuel Macron’s two five-year terms, to describe what the next president will inherit. It does not grade, does not judge and attributes no change to a policy, except when an official source does so: the source is then cited.

  • Start and end. The starting point is 2016, the last full year before the election of May 2017 (2015 for PISA, the last edition before 2017). The latest value is the most recent published at the data cut-off date; a more recent quarterly or monthly figure is given separately. The change is read in points for a rate or a share, and as a percentage for a level.
  • Sources. Public bodies only: INSEE, Eurostat, Banque de France, DG Trésor, finance acts, Cour des comptes, Haut Conseil des finances publiques, DREES, DEPP, Dares, France Travail, DGFiP, SSMSI, Citepa and SDES, RTE, Agreste, Agence Bio, Cerema, the ministries of the Armed Forces and of Foreign Affairs, OECD, NATO, the European Commission and the Council of the European Union, and courts for their decisions. Each series carries its body, its publication, its date and its series identifier or an exact excerpt. The national accounts are revised every year: the page always takes the latest vintage, and the whole series from the same vintage where possible.
  • Comparison. When Eurostat publishes the same series for the euro area or the European Union, it is shown in grey on the chart, over the same years.
  • Breaks and context. A change of method or source is flagged under the indicator. The 2020-2021 health crisis and the 2022 energy price shock affect almost all the series: they are recalled at the top of the page, without drawing any conclusion from them.
  • Choice of indicators. The main public series in each field (public finances, economy, employment, income and poverty, tax fraud, tax and social expenditures, the Paris 2024 Games, health and schools, defence, foreign affairs, energy, climate, environment and agriculture, housing, security), whether they have risen or fallen; in each field, the reference series comes first. When there is no complete official measure, as for tax fraud and evasion, the published estimates are cited with their author, date and method, and those that do not come from a public body are flagged as unofficial. Totals over several years are sums calculated by the editors from the published annual amounts, in current euros.
  • Dated facts. Power station closures, programming acts, military withdrawals, court decisions: each fact carries its exact date and the official source that establishes it (Journal officiel, decision of the court, press release from the Élysée or the ministry).
  • Paris 2024 Games. No annual series: the cumulative public cost and its breakdown as presented by the Cour des comptes in its reports to Parliament, the accounts of the organising committee (named as the source), and the dates of the organisation. A study commissioned by an organiser is cited with the name of whoever commissioned it.
  • Quotes. The warnings of the Cour des comptes and of the Haut Conseil des finances publiques and the statements of public officials are copied word for word from their source, dated and attributed.
  • The series, the dated facts and the quotes can be downloaded from the Data page (CSV and JSON, in French).

Explainers

Explainers set out facts sourced to the primary source and attributed arguments, without any conclusion from the editors.

The Explainers section publishes pages signed “the editors”. They set out facts and arguments; they do not conclude. The same rules apply to all of them.

  • Facts sourced to the primary source. Each figure refers, through a numbered note, to the body that produces it (INSEE, Cour des comptes, Haut Conseil des finances publiques, finance act…), with the title of the publication, its date, its address and an exact excerpt. A figure taken from an article or a third party is traced back to its source, or is not published.
  • Attributed arguments. The “What is being debated” part reports the arguments put forward, each attributed to whoever makes it (institution, organisation, public official, economist), with its source. No argument is presented without an author.
  • No conclusion from the editors. A page does not decide, recommends nothing and grades no one. A final, optional part states what we do not know.
  • Institutions and practices, never a candidate. A page covers an institution, a rule or a public practice, not a candidate. The same rules apply to all subjects, whatever the political side concerned.
  • Right of reply. A person, an institution or an organisation cited or implicated may send their reply through the contact form (subject “Right of reply or correction of an Explainer”).
  • Published corrections. An error corrected after publication is dated and described at the top of the page, which carries its update date, and listed in the errata.

Limitations

What our costings do not say: the cost schedule covers 85 levers out of 118, its sources date from 1986 to 2026, its ranges reflect the gap between sources, not statistical uncertainty.

Our costings are public, contestable estimates, not truths. What they do not say is summarised in the warning.

  • The cost schedule covers 85 levers out of 118, from 308 sources. The other levers (institutions, rules, targets) are compared without being costed.
  • The costs come from sources of different dates (from 1986 to 2026), in euros of their year: they are annual orders of magnitude, not yet recalibrated on a single price and GDP framework. The year of each coefficient is shown.
  • Several coefficients are derived from a budget total (1% of the cost of a benefit, for example), from a published yield, or brought down to the unit from a broader costing: the cost schedule flags this each time. Beyond the range measured by the source, an extrapolated amount (*) assumes that the cost remains proportional, which the source does not verify.
  • The ranges reflect the gap between sources, not statistical uncertainty. Adding ranges together gives an order of magnitude, not a confidence interval.
  • 9 of 770 positions remain to be reviewed; each cell says so (*).

What the calibration on 2022 shows

Applied to the 2022 programmes, the cost schedule costs only 6 to 16% of their published amounts, and its unit costs differ by a median 30% from those of the think tanks; what has been corrected, what remains open.

We costed, with today’s cost schedule, the 2022 programmes of Emmanuel Macron and Jean-Luc Mélenchon, then compared the result, line by line, with the costings published at the time by the Institut Montaigne and the Fondation iFRAP. The first finding is a limitation: the cost schedule fully costs only a small part of these programmes, from 6 to 16% of the published amounts depending on the programme and the think tank. This is why the site publishes no total as long as coverage remains low.

On the same assumption, across 23 unit costs tested, the median gap with the think tank is 30% in absolute value (38% before the corrections of 13 October 2026), and 11 tests fall below 25% (10 before the “net” convention for the index point, on 18 October 2026, which brings the central cost of one point down to exclude internal employer contributions and brings the cost of a 10% increase in the point excluding education closer to the iFRAP’s: €11.2bn against €14bn, instead of €18.6bn). The costs match for the retirement age, the index point, teachers’ pay, the marital quotient, the taxation of capital income and VAT. Elsewhere, they diverge markedly, and not in a single direction: our cost for an increase in the SMIC is nearly three times that of the Institut Montaigne (€7.6bn against €2.7bn for the increase promised in 2022), our costs for childcare places and free university are 76 to 87% higher, while we underestimate by more than half the revenue from making employee savings schemes subject to contributions, and by nearly half the cost of an increase in the minimum pension.

Each source concerned has been reread. Where it allowed, the cost schedule has been corrected: the cost of an additional police officer no longer counts unsourced equipment, two options costed against the law of 2022 (minimum pension at €1,400, minimum old-age income at €1,100) have been removed, and second-hand figures have been replaced by their primary source wherever it was found. When the correction would have mixed two scopes, it was not made: the gap on the SMIC arises because the Institut Montaigne deducts income tax and benefits and reasons on the reductions in social contributions of 2017, and the gap on employee savings because our source counts only social security contributions. These gaps remain open; they are written into the limitations of the lever concerned, and the cost of an increase in the SMIC, which several candidates cost, should be read with this caveat.

Finally, several 2022 measures have since become law (retirement age, minimum pension, abolition of the licence fee, CVAE): against the law of 2026, they no longer cost anything, and the cost schedule costs them at zero.

The cost schedule, lever by lever

Current law, the coefficients, the horizon, the limitations and the sources read, for each of the 118 levers of the cost schedule.

For each lever: current law, the coefficients and their range, the horizon, the limitations and the sources, with the excerpt read (sources in French). Amounts in billions of euros per year.

Income and work

Wages and SMIC (minimum wage)

Net monthly SMIC (minimum wage)costable, spending
  • Net monthly SMIC (minimum wage), full time. For +€100 in net monthly SMIC (about +6.8%): a cost of €3.75bn per year (2.5 to 5) (coefficient calculated by the editors from legisocial, Montant du SMIC 2026). Current value: 1,477.93 €/month. Range measured by the source: 1,477.93 to 1,691 €/month; beyond it, linear extrapolation is flagged. Range measured by the sources: from the current value to +14.4% (increase costed by Clément Carbonnier, SMIC raised to €1,600 net in 2024, i.e. about €1,691 at 2026 prices); no source read costs a decrease. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 9 sources

Current law. Gross hourly SMIC (minimum wage) €12.31 since 1 June 2026, i.e. €1,867.02 gross per month and about €1,477.93 net for 35 hours. This is an automatic 2.41% uprating due to inflation (Labour Code, Art. L3231-5). The net figure is calculated by legisocial with 20.84% in contributions, and the June text is not cited. On 1 January 2026, the SMIC was €12.02 gross, i.e. €1,443.11 net (Decree No. 2025-1228 of 17 December 2025, info.gouv.fr).

Horizon. Full year, €bn per year, in 2019 euros (lower bound) and 2024 euros (upper bound). The cost comes mainly from the general reductions in social contributions indexed to the SMIC, net of additional contributions and of the decrease in the in-work benefit (prime d'activité). The lower bound (expert group, 2019) also counts the increase in the public sector wage bill and the feedback from income tax and benefits; the upper bound (Carbonnier, 2024) counts neither the public sector wage bill nor the income tax feedback.

Limitations. Range between the two static costings read, with different scopes and legislation; neither relates to the 2026 RGDU (single general reduction in contributions). Lower bound, €2.5bn for +€100 of net SMIC: expert group on the SMIC, 2019 report (2019 legislation, table VI.1), effect on the general government balance after cushioning by the tax-benefit system, including €1.8bn of increase in the public sector wage bill. Upper bound, €5bn: costing by Clément Carbonnier for +€201 net (SMIC at €1,600, +14.4%, 2024 legislation), about €10bn of net cost before income tax and VAT feedback (€21bn of additional reductions, less €5bn of employee contributions, €3bn of employer contributions and €2.5bn of in-work benefit), scaled down linearly to €100; this costing does not count the public sector wage bill. Central value, €3.75bn: midpoint of these two costings, in the absence of a costing under the 2026 RGDU (coefficient calculated by the editors). The cost comes mainly from reductions in social contributions (lower revenue): it is classified as a revenue decrease. A SMIC increase also raises the lowest public sector salaries: declared overlap with the public sector pay index point. Outside the range, with a different scope: the « effet prix » [“price effect”] alone from the Institut Montaigne (€3.5bn, i.e. 1.75 per €100) and its updated costing of €19bn (17 to 21), which includes medium-term effects (employment, wage spillover) and is not static. Gaps between sources for the same measure: government €3bn, Rabault (PS) €23bn over three years, iFRAP +€32.5bn of reductions. In 2022, for a SMIC at €1,400 net, Montaigne costed €2.7bn. 2017 benchmark, under different law (general reduction, CICE, responsibility pact): the expert group costed a 1% increase in the gross SMIC at €0.24bn of deterioration of the general government balance, after deducting employee contributions, income tax and benefits; transposed to +€100 in 2026, this rule would give about €1.6bn (€2.4bn for the 2019 rule). The expert group's 2025 report does not cost a SMIC increase under the RGDU. Calibration on 2022: for the 2022 increase (+10.3%), a coefficient of 5 gave €7.6bn, against 2.7 at Montaigne. The cost depends heavily on the indexation of the reduction thresholds (2026 RGDU): a freeze of the thresholds in euros largely cancels it. Job losses estimated by Montaigne: 230,000 to 380,000 (not counted).

Wage increases and indexationnot costable
Current law, horizon, limitations and 1 source

Current law. No automatic indexation of private-sector wages to prices (only the SMIC is indexed, Labour Code, Art. L3231-4 et seq.). In the civil service, the value of the public sector pay index point is set by decree.

Horizon. Full year, €bn per year, in 2022 euros, not updated for 2032

Limitations. Not costed here: the cost of the civil service pay index point is carried by the « fonction-publique » lever. Indexation of private sector wages (sliding scale) has no direct static budgetary cost, except through public employers and indexed benefits: not costed. Benchmark: iFRAP 2022 costs « 2 points » [“2 points”] of index point value at €4bn (the think tank's own costing), consistent with the official cost of one point, €2.17bn to €2.27bn (Senate, 2022 and 2023 increases), carried by the « fonction-publique » lever.

Reductions in social contributions on low wagescostable, revenue
  • Annual reduction in social contributions, as an amount. For €1bn a year less in social contributions: a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
  • Increase in the net wage of private-sector employees paid up to 3 times the SMIC, through lower contributions (labour cost unchanged). For +1% of the net wage of each private-sector employee paid up to 3 times the SMIC (net wage bill concerned of about €385bn in 2024: 1% of that bill): a cost of €3.85bn per year (3.75 to 3.96) (coefficient calculated by the editors from Groupe d'experts SMIC, Rapport 2025 (25 novembre 2025)). Current value: 0 % du salaire net. Range measured by the source: 0 to 14 % du salaire net; beyond it, linear extrapolation is flagged. Arithmetic coefficient (1% of the net wage bill concerned, editors' calculation). Beyond about 14% of the net wage, employee contributions excluding CSG and CRDS (11.31% of gross pay up to the Social Security ceiling) are no longer sufficient: the measure must affect the CSG or employer contributions; the static cost follows the same rule, line flagged as extrapolated.
  • Increase in the net wage of all private-sector employees, through lower contributions (labour cost unchanged). For +1% of the net wage of each private-sector employee (net wage bill of about €551bn in 2024: 1% of that bill): a cost of €5.51bn per year (coefficient calculated by the editors from Groupe d'experts SMIC, Rapport 2025 (25 novembre 2025)). Current value: 0 % du salaire net. Range measured by the source: 0 to 14 % du salaire net; beyond it, linear extrapolation is flagged. Arithmetic coefficient (1% of the net wage bill, editors' calculation). Beyond about 14% of the net wage, employee contributions excluding CSG and CRDS are no longer sufficient: the measure must affect the CSG or employer contributions; the static cost follows the same rule, line flagged as extrapolated.
  • Uniform net gain per private-sector employee, through lower contributions (same amount for each). For +€1 net per month for each of the 20.96 million private-sector salaried jobs (Insee, second quarter of 2026): 12 × 20.957 million, i.e. €0.2515bn per year (€25.15bn for €100 per month): a cost of €0.252bn per year (coefficient calculated by the editors from Groupe d'experts SMIC, Rapport 2025 (25 novembre 2025)). Current value: 0 €/mois par salarié. Range measured by the source: from 0 €/mois par salarié; beyond it, linear extrapolation is flagged. Arithmetic coefficient (amount × 12 months × private-sector salaried jobs, editors' calculation), with no measured bound. At the SMIC, employee contributions excluding CSG and CRDS are worth about €211 per month (11.31% of €1,867 gross): a larger gain requires also cutting employer contributions and raising gross pay by the same amount; the static cost is the same.
  • Status quo of the 2026 RGDU (already in force): a gain of €0bn per year.
  • Complete abolition (static gross gain, 2026 total): a gain of €72.3bn per year.
  • Abolition of the general reductions paid above 1.6 times the SMIC (exit point brought down from 3 to 1.6 times the SMIC, reduction unchanged below): a gain of €14.3bn per year (12.7 to 15.8) (coefficient calculated by the editors from Groupe d'experts SMIC, Rapport 2025 (25 novembre 2025)).
Current law, horizon, limitations and 10 sources

Current law. Since 1 January 2026, the single tapered general reduction (RGDU, decree of 4 September 2025) merges the general reduction and the health and family contribution reductions (bandeaux maladie et famille), with an exit point at 3 times the SMIC. Cost of the general reductions: €75.4bn in 2024, €72.4bn forecast for 2025, €72.3bn in 2026 (Commission des comptes de la Sécurité sociale (Social Security Accounts Committee), October 2025).

Horizon. Full year, €bn per year; positive sign = gain for public finances unless marked « coût » [“cost”]. Net wage increase parameters: 2024 euros (wage bill) and jobs in the second quarter of 2026.

Limitations. The Expert Group's scenarios at 2 and 2.5 SMIC (exit at 2 SMIC: about −€6bn; Bozio-Wasmer: +€1.5bn) are costed « une fois les effets emploi pris en compte » [“once employment effects are taken into account”]: these are dynamic costings, cited here as benchmarks and removed from the options of the static cost schedule. Total abolition: 1.3 to 1.9 million jobs destroyed according to the Expert Group. Budget amendments also cite « près de 80 Md€ » [“nearly €80bn”] for all exemptions, a broader scope. Total abolition net of job losses (€40bn to €50bn) is a dynamic costing: it is not an option of the static cost schedule. A cut in contributions announced in euros per year is costed through the parameter in €bn per year: €1 less in contributions = €1 less in revenue, static, before any financing announced elsewhere. The additional income tax resulting from a rise in net wages is not counted. Before 2026, the general reduction (up to 1.6 SMIC) cost €39.4bn and the reduced health and family rates, which went beyond 1.6 SMIC, €36bn (2024, PLFSS 2026, annex 4). Since 2026, the single reduction runs up to 3 SMIC: no source read costs the share paid above 1.6 SMIC under the RGDU. Under the 2024 regime, chart I.24 of the SMIC Expert Group (2025 report) allows it to be estimated at about €15.8bn (reduced health rates €10.8bn and family rates €5.0bn paid above 1.6 SMIC, sum of the labels read by the editors). Option « suppression au-delà de 1,6 SMIC » [“abolition above 1.6 SMIC”] (editors' calculation): €15.8bn as upper bound (2024 regime); €12.7bn as lower bound, if the entire decrease in the cost of the reductions between 2024 and 2026 (€75.4bn → €72.3bn) fell on wages above 1.6 SMIC; midpoint as central value. Parameter in €bn: one euro less in contributions is one euro less in revenue; the additional corporate income tax that mechanically results is not counted (static). Net wage increases defined per employee (rule on public bases, 18 October 2026): same method for all, cost = per-person rule × public base, static, without the additional income tax. Private-sector gross wage bill paid up to 3 SMIC (2024, editors' calculation): €500bn based on the reduced family rate (1.8 points, €9.0bn paid between 1 and 3 SMIC, chart I.24), €494bn combining the reduced health rate (6 points, €26.3bn up to 2.5 SMIC, i.e. €438bn) and the family rate between 2.5 and 3 SMIC (€1.0bn, i.e. €56bn); net wage bill = gross wage bill × net/gross ratio, from 0.7588 (private-sector average, Insee 2024) to 0.7916 (employee deductions of 20.84% for a non-managerial employee up to the ceiling, Urssaf 2026): €374bn to €396bn, 385 as central value. Net wage bill of all private-sector employees: €726bn (Urssaf, 2024) × 0.7588 = €551bn. Uniform gain per employee: 20,957 thousand private-sector salaried jobs (Insee, second quarter of 2026; jobs, not people, including agriculture, temporary work and private household employers). Public employees are not counted (reading « salariés » [“employees”] as private-sector employees, general scheme contributions): if they were, about €2.0bn more per point of net wage (€244.6bn of gross salaries in 2024 × 0.81) or €7.3bn more per €100 per month (6.06 million public jobs).

In-work benefit (prime d'activité)costable, spending
  • Increase in the whole scale of the in-work benefit (prime d'activité). For +1% on the whole scale (flat amount and bonus): a cost of €0.125bn per year (coefficient calculated by the editors from Sénat, rapport d'information n° 728 (2025-2026) de MM. A. Bazin et P. Barros (commission des finances) sur l'enquête de la Cour des comptes relative à la prime d'activité (pages 6, 7 et 38 du PDF)). Current value: 0%. Range measured by the source: -100% to 17%; beyond it, linear extrapolation is flagged. Coefficient proportional to spending (editors' calculation). The largest increase measured by a source read is that of the bonus voted for 2026: €1.8bn in a full year for €10.7bn of spending, about +17%. Beyond that, the entry of new households is not counted: linear extrapolation, flagged as such.
Current law, horizon, limitations and 7 sources

Current law. Flat-rate amount of €638.28 for a single person since 1 April 2026 (+0.8%). Finance Act No. 2026-103 for 2026 and Decree No. 2026-222 of 30 March 2026 add an exceptional increase in the bonus (bonification), up to €54 per month, €50 on average. Spending: €10.41bn paid by the Caf (family allowance funds) in 2025 (10.3 in 2024), €10.7bn of actual spending in 2025 across all schemes (the Caf pay 97.2% of the total); the increase in the bonus costs €0.7bn in 2026 and €1.8bn in a full year (Senate, June 2026). 4.81 million recipient households at end-2024 (Drees, all schemes), 4.73 million in June 2026.

Horizon. Full year, €bn per year, in 2026 euros

Limitations. Derived coefficient: 1% of spending in a full year under 2026 law, i.e. €10.7bn spent in 2025 plus €1.8bn of increase in the bonus in a full year (Senate, report no. 728, June 2026), about €12.5bn; not counting new entrants. A single source for this base: no range. The Caf spending in 2025 (€10.41bn, Cnaf, PA Conjoncture no. 36), divided by their share of recipients across all schemes (97.2% in 2025, PA Conjoncture no. 38), gives about €10.7bn, the value used (check of 18 October 2026). Benchmark: +€10 per month per household for 4.81 million households (Drees, end-2024), i.e. about €0.58bn per year (arithmetic). Amendment CF1058 costs +3% at €540m on an erroneous base (« 18 milliards » [“18 billion”] for action 11 and an average amount of €327): not used. Increase in the bonus voted in the finance act for 2026 (€50 on average): €0.7bn in 2026, €1.8bn in a full year (Senate, report no. 728, figure taken from the Cour des comptes annex); the OFCE expects about €1bn more in 2027. Scope gap between sources: €10.3bn (Caf, 2024), €10.6bn (Cour des comptes, all schemes) and €10.7bn (spending executed in 2025); the initial PLF 2026 forecast €9.3bn.

Profit sharing (participation, intéressement)costable, revenue
  • Abolition of the income tax exemption (tax expenditure 120108): a gain of €2.89bn per year.
  • New profit-sharing or incentive scheme obligations (no static budgetary cost): a gain of €0bn per year.
  • Make profit-sharing, incentive schemes, company savings plans and employee share ownership subject to standard contributions (base exemptions, 2026): a gain of €3.7bn per year (3.7 to 4.9).
  • Exempt sums paid under profit-sharing (participation) from the forfait social (20%): a cost of €2.42bn per year (coefficient calculated by the editors from PLFSS 2026, annexe 4, tableau 9, Coût total des exemptions d'assiette, et de Legifiscal, Forfait social et fiscalité de la participation (mis à jour le 9 février 2024)).
  • Abolish the employer contribution on free share awards and stock options (30%): a cost of €1.5bn per year (1.5 to 1.56) (coefficient calculated by the editors from PLFSS 2026, annexe 4, tableau 9, Coût total des exemptions d'assiette).
Current law, horizon, limitations and 5 sources

Current law. Sums paid under participation, intéressement, employer matching (abondement) and the value-sharing bonus are exempt from income tax (CGI (General Tax Code), Art. 81-18°, 163 bis AA, 163 bis B, etc.; tax expenditure 120108). PPV (value-sharing bonus) paid: €3.4bn in 2024, compared with €5.2bn in 2023.

Horizon. Full year, €bn per year, in 2025 euros

Limitations. The tax cost is an « ordre de grandeur » [“order of magnitude”] (reliability indicated by the DGFiP), constant at €2.75bn from 2023 to 2025. Making profit-sharing (participation) compulsory in more companies has no direct static budgetary cost: effect on corporate income tax not costed. Social component (PLFSS 2026, annex 4, table 9): subjecting participation, intéressement, the PEE, stock options and free shares to ordinary social contributions would raise €3.7bn net in 2026 (central), after deducting the forfait social and specific contributions; €4.9bn gross, excluding general reductions (upper bound). 2026 forecast before the PLFSS 2026 measures. The PERCO and the collective company PER (€0.2bn net) are classified separately, with supplementary pensions. Known limitation: the PLFSS annex counts only social security contributions; supplementary pension and unemployment insurance contributions, which are also general government revenue, are missing, so the option underestimates the gain. The Institut Montaigne (2022) costs the same measure, including all ordinary contributions and after deducting the forfait social, at €9bn (base of €20.3bn in 2019, static); this costing, with a different scope and year, does not enter the range. No source read gives it for 2026. The income tax exemption is a tax expenditure: declared overlap with the across-the-board trimming of tax expenditures. The total yield of the forfait social and employer contributions on employee share ownership is not published by scheme; only the 2025 increase on free shares (from 20 to 30%) is costed, at €0.5bn. The PLF 2026 costs the income tax exemption (120108) at €2.89bn in 2024, 2025 and 2026, against €2.75bn in the PLF 2025: the « suppression-exoneration-ir » option uses €2.89bn. Rule on public bases (18 October 2026), editors' calculations: exempting participation from the forfait social costs at most 20% × €12.1bn of participation (exempted base forecast for 2026, PLFSS 2026, annex 4), i.e. €2.42bn; the figure overestimates the cost, as participation paid by companies with fewer than 50 employees, already exempt, is not isolated. Abolishing the employer contribution on free share awards and stock options (30% since 2025) costs about €1.5bn: the rate increase from 20 to 30% brings in €0.5bn, i.e. €0.05bn per point (lower bound and central value); 30% of the €5.2bn base for 2026 gives €1.56bn (upper bound). The forfait social on employer matching (abondement) in securities is not costed.

  • PLF 2025, Évaluation des voies et moyens, tome II, 2024-10: « 120108 – Exonération des sommes versées au titre de la participation, de l'intéressement […] 2 750 / 2 750 / 2 750 ; Fiabilité du réalisé : Ordre de grandeur »
  • Groupe d'experts SMIC, Rapport 2025, 2025-11-25: « le montant total versé est passé de 5,2 milliards d'euros à 3,4 milliards d'euros entre ces deux années »
  • PLFSS 2026, annexe 4, tableau 9, Coût total des exemptions d'assiette, 2025-10: « I. Participation financière et actionnariat salarié : 2026 (p), assiette exemptée 32,3 ; cotisations équivalentes brutes 4,9 ; nettes 3,7 (en Md€). Dont participation aux résultats de l'entreprise 12,1 / 1,9 / 1,3 ; intéressement 12,4 / 2,4 / 1,8 ; plan d'épargne en entreprises (PEE) 2,5 / 0,6 / 0,6 ; stock options et attributions gratuites d'actions 5,2 / 0,0 / 0,0. ; Pour cette estimation de perte de cotisations et contributions sociales, il est également tenu compte des contributions spécifiques (tel que le forfait social) qui sont prélevées sur les sommes exclues de l'assiette des cotisations. La perte de cotisations nette correspond à l'effet des exemptions d'assiette sur les cotisations de droit commun. A titre d'information, la perte de cotisations brute, qui résulte des exemptions d'assiette hors dispositifs d'allégements généraux est aussi estimée. ; La mesure vise à relever la contribution patronale spécifique due sur les attributions gratuites d'actions pour un surcroit de recettes, de l'ordre de 0,5 Md€ (hors changement de comportement) […] par le relèvement du taux de la contribution patronale spécifique applicable sur les AGA à 30 % »
  • PLF 2026, Évaluation des voies et moyens, tome II, Dépenses fiscales, 2025-10: « 120108 – Exonération des sommes versées au titre de la participation, de l'intéressement, de l'abondement, d'un partage de plus-value, des primes de partage de la valeur […] Réalisation 2024 : 2 890 ; Prévision 2025 : 2 890 ; Prévision 2026 : 2 890 ; Fiabilité du réalisé : Ordre de grandeur »
  • Legifiscal, Forfait social et fiscalité de la participation (mis à jour le 9 février 2024), 2024-02-09: « La participation est soumise au forfait social au taux de 20%. […] Seules les entreprises de moins de 50 salariés sont exonérées de ce forfait. »

Unemployment

Unemployment, work period required to qualify for benefitscostable, spending
  • Work period required to open entitlement. For 1 month less of required affiliation (between 6 and 4 months), for all new claimants (excluding first-time entrants, already at 5 months): a cost of €0.4bn per year (0.3 to 0.5) (coefficient calculated by the editors from Unédic, « Impact de l'évolution des règles d'assurance chômage au 1er juillet 2021 » (estimations de mars 2021), page 7 du PDF). Current value: 6 months. Range measured by the source: 4 to 6 months; beyond it, linear extrapolation is flagged. Range measured by Unédic: move from 4 to 6 months (estimate of March 2021). Outside this interval, linear extrapolation, flagged as such.
Current law, horizon, limitations and 4 sources

Current law. Qualifying for benefits requires 6 months of work (130 days or 910 hours) in the last 24 months, 36 months from age 55. This rule comes from Decree No. 2019-797 of 26 July 2019 and was renewed by the agreement of 15 November 2024. Before 2019: 4 months out of 28. Since 1 April 2026, first-time claimants (people who have received no benefit in the past 20 years) qualify with 5 months of work (108 days or 758 hours) in the last 24 or 36 months (Act No. 2025-989 of 24 October 2025, provisions approved by order of 28 March 2026): €100m in 2026, €130m per year once fully phased in from 2027 (Unédic).

Horizon. Full year, €bn per year, in 2021 euros

Limitations. Derived from the Unédic estimate of March 2021: €0.8bn per year at steady state for raising the qualifying-employment condition from 4 to 6 months, divided by two months. Returning to 4 months out of 28 costs about €0.8bn. The 2021 figure is not updated: unemployment and the SMIC have changed since, and the 2023 countercyclicality reduces the duration of entitlements. Benchmark under 2026 law: lowering the condition by one month for first-time claimants only costs €0.13bn per year (Unédic), much less than the coefficient, which applies to all entrants. « Net » convention: Unédic presents its reduced spending for 2021 and 2023 together with the funding of claimants' supplementary pensions (a flow to Agirc-Arrco, between general government bodies); for the 2019-2021 reform, the share of this flow is not isolated and could not be removed.

Unemployment, max. benefit durationcostable, spending
  • Maximum benefit duration before age 55. For 1 more month of maximum duration (between 18 and 24 months, under 55): a cost of €0.7bn per year (0.7 to 0.75) (coefficient calculated by the editors from Unédic, Réforme 2023 : premiers effets de la réforme de contracyclicité). Current value: 18 months. Range measured by the source: 18 to 24 months; beyond it, linear extrapolation is flagged. Range measured by Unédic: gap between 18 and 24 months (2023 countercyclicality reform). Outside this interval, linear extrapolation, flagged as such.
  • Abolition of countercyclicality (return to 24 months): a cost of €4.2bn per year (4.2 to 4.5).
  • Cancellation of the amendment on negotiated terminations (return to 18 and 22.5 or 27 months): a cost of €0.94bn per year (0.72 to 0.945).
Current law, horizon, limitations and 5 sources

Current law. Maximum duration: 18 months under 55 since countercyclical modulation (coefficient 0.75 since 1 February 2023). Older workers: 22.5 months at 55 and 56, 27 months from 57 (agreement of 15 November 2024). After a negotiated termination (rupture conventionnelle) concluded on or after 1 September 2026: 15 months under 55, 20.5 months from 55 (Act No. 2026-470 of 11 June 2026, order of 19 June 2026).

Horizon. Full year, steady state (around 2027-2029), €bn per year

Limitations. The linear coefficient is calculated by the editors: €4.2bn of benefits (Unédic, February 2023) divided by 6 months, i.e. €0.7bn per month; 0.75 as upper bound with the funding of supplementary pensions. « Net » convention: the €4.5bn saving includes €0.3bn that Unédic pays to the supplementary pension funds (Agirc-Arrco), a flow between general government bodies; the central value (€4.2bn) excludes it, the upper bound (4.5) includes it. The true profile is not linear: the last months cost less, as many claimants return to work before their entitlements end. The €4.5bn saving assumes « une conjoncture stable » [“stable economic conditions”]: if unemployment exceeds 9%, the duration automatically goes back up. Amendment on negotiated terminations (ruptures conventionnelles): €720m to €945m per year according to Unédic, €940m once the reform is fully in force (Senate report, the Unédic note not being published). Context: in 2026, the scheme's spending (€46.6bn) exceeds its revenue (€44.3bn), i.e. a balance of −€2.3bn (Unédic forecast of 17 June 2026; €46.3bn and €44.2bn in the March forecast). The options « suppression de la contracyclicité » [“abolition of countercyclicality”] (2023 reform) and « abrogation de la réforme 2019-2021 » [“repeal of the 2019-2021 reform”] (amount lever) relate to distinct reforms: the 2023 one was costed on the law resulting from the 2021 one, and Unédic adds the two together (€6.7bn at steady state, February 2023, table A); declared overlap, with zero common base: the two add up for the same candidate. The duration parameter and the option abolishing countercyclicality relate to the same mechanism (declared overlap). The June 2024 rules (a decree never applied) were costed by Unédic against the law of the time, since amended (agreement of 15 November 2024, 2026 amendment): no costing read relative to current law.

Benefit amount and taperingcostable, spending
  • Abolition of the tapering of high benefits: a cost of €0.46bn per year.
  • Return to the pre-2021 calculation of the reference wage: a cost of €1bn per year.
  • Repeal of the whole 2019-2021 reform (access, calculation, tapering): a cost of €2.26bn per year.
Current law, horizon, limitations and 2 sources

Current law. Since the 2019-2021 reform (Decree No. 2019-797 as amended), the benefit is calculated on a daily reference wage established over 24 months, including periods not worked. 30% tapering after 6 months for those with high previous earnings, with older workers aged 57 and over exempted since the 2024 agreement.

Horizon. Full year, €bn per year, in 2021 euros (steady state 2023-2024)

Limitations. Unédic estimate of March 2021 (steady state), not updated: €1bn for the calculation of the reference wage, €0.8bn for the qualifying-employment condition, €0.46bn for degressivity, €2.26bn in total. No sourced coefficient for « +1 point de taux de remplacement » [“+1 point of replacement rate”]. The OFCE table for the 2024 legislative elections gives €4bn for the « annulation de la réforme » [“cancellation of the reform”], without stating the author or the scope. Unédic presents the reduced spending of the 2021 and 2023 reforms « d'allocations chômage (et financement des caisses de retraite complémentaire) » [“in unemployment benefits (and funding of supplementary pension funds)”] (February 2023, table A): the share paid to supplementary pensions, a flow between general government bodies, is not isolated for the 2019-2021 reform and could not be removed. Repealing 2019-2021 overlaps with the abolition of countercyclicality (duration lever) and the access lever: declared overlaps.

Obligations of jobseekersnot costable
Current law, horizon, limitations and 1 source

Current law. Act No. 2023-1196 of 18 December 2023 for full employment: since 1 January 2025, all RSA (minimum income) recipients are automatically registered with France Travail and sign a commitment contract. The amendment of 25 February 2026 provides for intensive follow-up after a negotiated termination (rupture conventionnelle).

Horizon. Not applicable.

Limitations. Budgetary effects are purely behavioural (exits, sanctions): excluded from static costing. The cost of support by France Travail has not been recorded.

  • service-public.gouv.fr, RSA (F502): « depuis le 1er janvier 2025, chaque bénéficiaire du RSA est inscrit automatiquement à France Travail et doit signer un contrat d'engagement »
Governance of unemployment insurancenot costable
Current law, horizon, limitations and 2 sources

Current law. Scheme managed by Unédic under an agreement between the social partners (agreement of 15 November 2024), approved by the State, which issues a framework letter and takes part of the scheme's revenue.

Horizon. Not applicable.

Limitations. Institutional question. Point to note: without the State's levies (€4.1bn), the scheme's balance would be +€1.8bn in 2026 instead of −€2.3bn (Unédic, forecast of 17 June 2026; +€2.0bn in the March forecast).

Working time and labour law

Statutory working timenot costable
Current law, horizon, limitations and 1 source

Current law. Statutory working time: 35 hours per week (Labour Code, Art. L3121-27), beyond which overtime applies.

Horizon. Not applicable.

Limitations. Statically, the statutory working week has no direct cost for public finances. The effects (employment, wages, cost of public hiring at 32 hours, overtime) are behavioural. For the 32-hour week in the civil service or a 6th week of paid leave, no official costing has been read. The 2024 OFCE table cites €1.6bn per year (Montaigne 2022) for the 6th week, a distinct measure. To be treated as a dynamic effect, separately.

Overtimecostable, revenue
  • Abolition of the income tax exemption (tax expenditure 120146): a gain of €2.25bn per year.
  • Abolition of the reduction in employee contributions (not offset): a gain of €2.31bn per year.
  • Abolition of both exemptions: a gain of €4.56bn per year.
  • Abolition of the annual overtime quota (220 hours in the absence of an agreement), no static budgetary cost: a cost of €0bn per year.
Current law, horizon, limitations and 5 sources

Current law. Statutory pay premium of 25% then 50% (Labour Code, Art. L3121-36, unless otherwise agreed). Income tax exemption up to €7,500 per year (CGI (General Tax Code), Art. 81 quater) and an 11.31% reduction in employee old-age contributions. An amendment abolishing the ceiling was adopted at first reading in the National Assembly on 25 October 2025: according to concurring sources, the ceiling remains in the final act (Journal officiel text not cited).

Horizon. Full year, €bn per year, in 2025-2026 euros

Limitations. The tax cost has « ordre de grandeur » [“order of magnitude”] reliability. Annex 2 of the PLFSS 2025, cited elsewhere, gives €2.485bn. Static: the likely fall in declared hours is not counted. Abolition of the €7,500 ceiling: the only costing found comes from an accounting firm (€1bn) and is not used. The €2.31bn cost of the reduction in employee contributions comes from annex 4 of the PLFSS 2026 (€2,306m forecast for 2026, not compensated). The flat-rate deduction of employer contributions for companies with fewer than 250 employees also costs €872m in 2026 (compensated). The PLF 2026 raises the cost of the income tax exemption (120146): €2,144m actual in 2024, €2,250m forecast for 2026, against €1,787m forecast for 2025 in the PLF 2025, after a change of method (+€0.3bn): the « fin-exoneration-ir » and « fin-deux-exonerations » options use €2.25bn and €4.56bn. Abolishing the annual overtime quota has no static cost: with unchanged hours, nothing changes for public finances; the effect works through additional hours, which the cost schedule does not count.

Sunday and night worknot costable
Current law, horizon, limitations and 1 source

Current law. Sunday rest as a principle (Labour Code, Art. L3132-3), with permanent, sector-specific and geographical exemptions (tourist zones, mayor's Sundays), widened by Act No. 2015-990 of 6 August 2015.

Horizon. Not applicable.

Limitations. No direct static budgetary effect.

Scale of employment tribunal (prud'hommes) awardsnot costable
Current law, horizon, limitations and 1 source

Current law. Mandatory scale of compensation for dismissal without real and serious cause (Labour Code, Art. L1235-3, Ordinance No. 2017-1387 of 22 September 2017).

Horizon. Not applicable.

Limitations. No budgetary effect. The scale ranges from 3 to 20 months of gross salary beyond 30 years of service.

Minimum incomes and benefits

RSA (minimum income), amount and conditionscostable, spending
  • Increase in the RSA (minimum income) flat-rate amount beyond indexation. For +1% on the RSA flat-rate amount (beyond indexation): a cost of €0.17bn per year (0.15 to 0.17) (coefficient calculated by the editors from solidarites.gouv.fr, Revalorisation annuelle au 1er avril 2026). Current value: 0%. Range measured by the source: 0% to 10%; beyond it, linear extrapolation is flagged. Range measured: cumulative increase of about 10% in the flat-rate amount (exceptional upratings from 2013 to 2017). Beyond that, the entry of new households is no longer covered: linear extrapolation, flagged as such. A decrease in the flat-rate amount, which no source read costs, is costed in the same way, by linear extrapolation flagged as such.
  • RSA paid automatically to all those entitled (non-take-up eliminated; DREES: €750m per quarter not paid in 2018): a cost of €3bn per year.
  • RSA payment limited to three years (end of entitlement beyond, for all recipients, without exemption; static effect): a gain of €6.3bn per year (4.5 to 7.2) (coefficient calculated by the editors from solidarites.gouv.fr, Revalorisation annuelle au 1er avril 2026).
Current law, horizon, limitations and 12 sources

Current law. Flat-rate amount of €651.69 per month for a single person with no resources since 1 April 2026 (+0.8%). Open from age 25, and from 18 to 24 for single parents or young people with a certain period of work. Automatic registration with France Travail and commitment contract since 1 January 2025 (Act No. 2023-1196). Spending: €12,218m in 2023, including the Christmas bonus (Drees), and €12.2bn paid by the Caf (family allowance funds) in 2025; 1.84 million recipient households at end-2024 (Drees), 1.89 million in June 2026.

Horizon. Full year, €bn per year

Limitations. Option versement-automatique-rsa: same costing as the recours-total-rsa option of the versement-automatique lever (DREES, Dossiers no. 92: €750m of RSA (minimum income) not paid per quarter in 2018, i.e. €3bn per year, 2018 euros, not updated), for a position filed under the RSA; the two overlap entirely (declared overlap). Option limite-trois-ans (calculation by the editors, 18 October 2026): RSA spending (€12.2bn paid by the Caf (family allowance funds) in 2025) × share of beneficiaries (recipients and spouses) present for at least three years, which DREES does not publish as such: at least 37% (five years or more, end of 2023: lower bound, €4.5bn), at most 59% (two years or more: upper bound, €7.2bn), about 52% by spreading the two-to-five-year band (22%) evenly: central value, €6.3bn. The share of beneficiaries is taken as the share of spending. Static: neither shifts to other benefits (accommodation, social assistance, other minimum benefits) nor returns to work are counted; an exemption (people unfit for work, on a social support pathway, or who fulfil a requirement in return) reduces the saving in an unmeasured proportion. The former coefficient (€0.13bn per % increase in the flat-rate amount, i.e. 1% of 2022 spending) is withdrawn: the allowance is differential, and raising the flat-rate amount by €X adds €X to each household whatever its resources. The 1.89 million households × €6.52 (1% of the flat-rate amount for a single person) × 12 months alone give €0.15bn per %, before the family scale and new entrants. The exact cost requires the distribution of households according to the family scale (DREES, CNAF) or a microsimulation (OpenFisca), not available here. The flat-rate amount of the in-work benefit (prime d'activité) is a separate parameter. Non-take-up (34%) reduces the actual cost. Coefficient used: the State set at €1.40bn per year, from September 2018, the cost borne by the départements due to the exceptional increases in the flat-rate amount from 2013 to 2017 (five decrees, about +10% in total excluding inflation), i.e. €0.14bn per point, in 2018 euros. This figure relates to the RSA alone (the in-work benefit is another lever) and includes households that entered thanks to the raising of the threshold. Updated to the 2026 flat-rate amount (€651.69 against €550.93 in 2018, i.e. +18%): 0.14 × 651.69 / 550.93 ≈ €0.17bn per point, central value and upper bound since 18 October 2026 (calculation by the editors, in 2026 euros); lower bound, €0.15bn, the mechanical effect alone on households already receiving the allowance (1.89 million × €6.52 × 12 months, DREES and 2026 order), before the family scale. The entry of new households opened up by a large increase in the flat-rate amount is included only within the limit of the 2013 to 2017 increases (about 10%): beyond that, it is not costed, and a position that goes further is partial. Full take-up (versement-automatique lever) would increase this cost. External benchmarks, not counted, for an income guarantee at the poverty line, a broader scope than an increase in the flat-rate amount: €36bn according to iFRAP (April 2022, threshold of €1,063, of which €9.5bn for young people); about €25bn according to La France insoumise (2021 amendments); €30bn in its counter-budget of November 2017.

Minimum income for under-25scostable, spending
  • Opening an RSA-type minimum, merged with the prime d'activité, to 21-24-year-olds (extra cost of the age criterion alone): a cost of €5.2bn per year.
  • Opening to 18-24-year-olds (21-24 + 18-20): a cost of €11.8bn per year (7.8 to 14) (coefficient calculated by the editors from service-public.gouv.fr, fiche F32700 « Contrat d’engagement jeune (CEJ) », vérifiée le 1er avril 2026).
  • Variant with housing benefits included, 18-24-year-olds: a cost of €14.2bn per year.
Current law, horizon, limitations and 3 sources

Current law. No general minimum social benefit before 25: the RSA (minimum income) is open to 18-24-year-olds only for single parents or young people with a sufficient period of work. The youth commitment contract (contrat d'engagement jeune) opens entitlement to an allowance of at most €566.17 per month for an adult who is independent for tax purposes or attached to a non-taxable household (€339.70 if each income share of the household falls in the first bracket of the scale; €226.48 for a minor), compared with €651.69 for the RSA of a single person (service-public, April 2026).

Horizon. Full year, €bn per year, at 2018 legislation and prices

Limitations. IPP simulations from 2018 of a « revenu de base » [“basic income”] merging the RSA and the in-work benefit, paid automatically (100% take-up), with a minimum of €461 per month, lower than the current RSA of €651.69. Low: 34% non-take-up applied (DREES), i.e. about 7.8. High: uprating in proportion to the RSA since 2018, i.e. about 14 (calculations by the editors, without microsimulation). The IPP points out that detaching young people from their parents' tax household partly reduces the cost (lower family quotient), already counted. The programmes' « garanties d'autonomie » [“autonomy guarantees”] (at the poverty line, or at 50% of the net SMIC, for young people detached from the parental tax household) have no equivalent in the cost schedule: the options simulate an income of €461 for all 18-24 year-olds. No static costing of these schemes has been read.

Family allowancescostable, spending
  • Increase in family allowances. For +1% on family allowances: a cost of €0.139bn per year (0.134 to 0.14) (coefficient calculated by the editors from solidarites.gouv.fr, Revalorisation annuelle au 1er avril 2026). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • End of income-related modulation (return to universal amounts): a cost of €1bn per year (0.75 to 1).
  • Family allowances from the first child, €70 a month, without modulation (DSS): a cost of €1.8bn per year (1.8 to 2.8).
  • Return of the age supplement at 14 (cancellation of the March 2026 postponement to 18): a cost of €1.28bn per year.
  • Family allowances reserved for families of French nationality (share of foreigners in the population applied to spending; upper bound pending from the Cnaf): a gain of €1.27bn per year (coefficient calculated by the editors from solidarites.gouv.fr, Revalorisation annuelle au 1er avril 2026).
  • Family income of €240 per month per minor child, from the first child, in place of family allowances (supplements included) and the complément familial (net cost of these two benefits; family quotient counted under its own lever): a cost of €23.4bn per year (23.4 to 23.9) (coefficient calculated by the editors from solidarites.gouv.fr, Revalorisation annuelle au 1er avril 2026).
Current law, horizon, limitations and 14 sources

Current law. Family allowances paid from the 2nd child: €153.01 per month for 2 children since 1 April 2026 (basic amount), adjusted according to income since 2015, uprated every 1 April in line with inflation over the previous twelve months. Age supplement postponed from 14 to 18 for children born on or after 1 March 2012 (Decree No. 2026-138, in force on 1 March 2026). Spending: €13,909m in 2025, all schemes (Commission des comptes de la sécurité sociale (Social Security Accounts Committee), October 2026), €13,711m forecast for 2026 and €13,407m for 2027; 4.69 million households receiving benefits from the Caf (family allowance funds) in December 2025.

Horizon. Full year, €bn per year, in 2025 euros for the coefficient; each option in the euros of its source

Limitations. Coefficient calculated by the editors: 1% of €13.909bn (2025, central value), of €13.407bn forecast for 2027 after the postponement of the age supplement (lower bound) and rounded to 0.14 as upper bound; the former coefficient (0.13, 1% of €13.04bn in 2022) was replaced on 18 October 2026. The text introducing means-testing (LFSS 2015) is not cited at source. End of means-testing: low = €750m per year (National Assembly fact-finding mission, 2020, report no. 3168, based on a 2016 Cnaf study, amount at the time); central and high = yield of means-testing estimated at about €1bn in 2025 by the Direction de la sécurité sociale (same report). The Senate costed abolition at €830m in 2022. From the first child: DSS costing for €70 per month without means-testing; central = net cost after the decrease in RSA and in-work benefit (€1.8bn), high = gross cost (€2.8bn); the National Assembly's LexImpact unit gives €3.58bn for €75 per month means-tested (different scope, not counted). A position that does not state the amount paid for the first child is not linked to this option. Age supplement: €1.28bn per year at full roll-out, reached in 2030 (€1.17bn in 2029), and €210m in the first year: costing by the Direction de la sécurité sociale on Cnaf data, taken up by the HCFEA (note adopted on 10 March 2026). The first three options are each costed all other things being equal: ending means-testing also applies to the age supplement, hence a slight overlap if both are adopted (declared overlap). Option reserve-francais (decision of 18 October 2026, same rule for all candidates who reserve a benefit for French nationals): lower bound, share of foreigners in the population (9.1% in 2025, Insee) × family allowance spending (€13,909m in 2025), i.e. €1.27bn; upper bound, the share of allowances paid to foreign recipients, which neither the Cnaf nor the Drees publishes: pending a reply from the Cnaf (request sent), it is taken as equal to the lower bound and the position is marked partial. Benchmarks not counted: at 31 December 2013, about 2% of family allowance beneficiaries were nationals of the European Economic Area or Switzerland and 6% of other States (headcounts, ministerial reply to written question no. 29143, Cnaf data); about 11% of the households receiving benefits from the Caf, all benefits combined, are of foreign nationality (Cnaf, December 2025, no reference year). As foreign families have more children on average, the share of spending may exceed the share of households. Reading: nationality of the recipient; a family in which one parent is French and the other is not could remain covered, which would reduce the saving. Legal compliance not assessed (principle of equality, European Union law, bilateral social security agreements), as elsewhere in the cost schedule. Option revenu-familial-240 (calculation by the editors, 18 October 2026): €240 × 12 months × 13,868,000 minor children living in families (Insee, 2023 census, France excluding Mayotte), i.e. €39.94bn gross, minus the two abolished benefits that the measure names and that fall under this lever: family allowances, age supplement included (€13,909m in 2025), and the complément familial (€2,593m), i.e. €23.4bn net; upper bound with the spending forecast for 2027 (€13,407m and €2,605m, CCSS), €23.9bn. The family quotient, which the measure also abolishes, is counted under the quotient-familial lever (€12.5bn to €16.5bn). Not counted: the condition of five years of professional activity and lawful residence, which excludes an unpublished number of families (the cost is therefore overestimated accordingly), the other merged schemes that the measure does not name, and the third-child bonus (separate measure). The candidate counts 14 million minors (€40.3bn).

Automatic payment, single social allowancecostable, spending
  • 100% take-up of the RSA and the prime d'activité (automatic payment, IPP): a cost of €2.9bn per year (2 to 3).
  • 100% take-up of the RSA alone (DREES: €750m per quarter not paid out in 2018): a cost of €3bn per year.
  • Simplification and merger of RSA + prime d'activité with no losers (on top of automatic payment): a cost of €1.5bn per year.
  • Merger of RSA, prime d'activité and housing benefits with no losers (on top of automatic payment): a cost of €8.6bn per year.
Current law, horizon, limitations and 3 sources

Current law. « solidarité à la source » (solidarity at source) reform: pre-filling of RSA (minimum income) income declarations, piloted in autumn 2024, with nationwide roll-out planned for the first quarter of 2025. The « allocation sociale unique » (single social allowance) project is not in force.

Horizon. Full year, €bn per year, in 2018 euros

Limitations. The IPP considers the €2.9bn figure uncertain (« approximation » [“approximation”]); the DREES measures 34% quarterly non-take-up of the RSA and 20% persistent non-take-up. Gap between sources: the IPP costs €2.9bn for the RSA and in-work benefit together, the DREES about €3bn for the RSA alone, on different fields and vintages. For the single social allowance, the cost depends entirely on its parameters: 0 if it is done at constant cost, with losers. The lower bound (2) is an assumption by the editors: partial elimination of non-take-up. A position that promises automatic payment of social benefits in general is linked to the option of 100% take-up of the RSA and in-work benefit, as a partial costing: other benefits are not costed. Benchmark not counted: the Institut Montaigne (2022) costs automatic payment of all benefits at €4.3bn to €7.9bn (6.1 as central value), but its range rests on an assumed non-take-up of 0 to 10% for housing benefits and family allowances, without source, it deducts €1.7bn of supposed fraud assumed to be entirely avoided, and it counts €6bn for the RSA and in-work benefit, double the IPP costing used here.

Universal incomecostable, spending
  • Means-tested basic income (merger of RSA + prime d'activité, automatic), aged 25 and over: a cost of €4.4bn per year.
  • Same, from age 21: a cost of €9.6bn per year.
  • Same, from age 18: a cost of €16.2bn per year.
  • Basic income including housing benefits, automatic, from age 18: a cost of €25.7bn per year.
Current law, horizon, limitations and 2 sources

Current law. No universal income. Minimum social benefits (€30.6bn in 2022) and the in-work benefit (prime d'activité) are means-tested and must be applied for.

Horizon. Full year, €bn per year, at 2018 legislation and prices

Limitations. These are tapered basic incomes (rate of 30% or 38%, minimum of €461 or €725), not unconditional universal incomes. For an unconditional income paid to all adults, no official costing has been read, and these options do not apply to it. The amounts are those of 2018, not updated to 2026 scales (RSA about +18% since 2018). These basic incomes include automatic payment and extension to young people: declared overlaps with the “Automatic payment, single social allowance” and “Minimum income for under-25s” levers.

  • IPP, Rapport n° 18 (2018), 2018-06: « COÛT TOTAL ADDITIONNEL : 4,4 milliards d'euros ; 9,6 ; 16,2 ; 25,7 milliards d'euros annuels »
  • DREES, Minima sociaux 2024, 2024-10: « En 2022, les dépenses liées au versement des minima sociaux s'élèvent à 30,6 milliards d'euros »

Early childhood and old age

Nurseries and childcarecostable, spending
  • Additional crèche places (funded by the single service benefit, PSU). For 10,000 additional crèche places: a cost of €0.2bn per year (coefficient calculated by the editors from DREES, Minima sociaux 2024, fiche 34, tableau 4). Current value: 0 places. Range measured by the source: 0 to 70,000 places; beyond it, linear extrapolation is flagged. Range measured by the Cour des comptes: 35,000 then 70,000 crèche places. Beyond that, linear extrapolation, flagged as such.
  • Additional early childcare places, all types (crèches and childminders). For 10,000 additional childcare places, all types: a cost of €0.15bn per year (0.13 to 0.15) (coefficient calculated by the editors from DREES, Minima sociaux 2024, fiche 34, tableau 4). Current value: 0 places. Range measured by the source: 0 to 200,000 places; beyond it, linear extrapolation is flagged. Range measured by the Cour des comptes: up to 200,000 additional places (2030 target). Beyond that, linear extrapolation, flagged as such.
  • Abolition of the childcare costs tax credit: a gain of €1.8bn per year (1.74 to 1.8).
  • +1% on the complément de libre choix du mode de garde (childcare choice supplement, CMG): a cost of €0.074bn per year (0.074 to 0.077) (coefficient calculated by the editors from DREES, Minima sociaux 2024, fiche 34, tableau 4).
Current law, horizon, limitations and 5 sources

Current law. Early childhood benefit (Prestation d'accueil du jeune enfant, Paje): €11,510m in 2025, including €7,372m for the childcare choice supplement (complément de libre choix du mode de garde) (Commission des comptes de la sécurité sociale (Social Security Accounts Committee), October 2026; the September 2025 reform of the supplement costs €0.5bn in a full year). Tax credit for childcare costs for children under 6 (CGI (General Tax Code), Art. 200 quater B): €1,737m actual in 2024, €1,799m forecast for 2025 and 2026, for 1.93 million households (PLF 2026). Funding of nurseries (Cnaf single service benefit, municipalities) and the public early childhood service are not cited in the source.

Horizon. Full year, €bn per year

Limitations. The +1% of the childcare supplement (complément mode de garde) is calculated by the editors: 1% of €7.372bn (2025), €7.69bn forecast for 2027 as upper bound (CCSS, October 2026); former coefficient 0.07 (1% of €6.76bn in 2022), replaced on 18 October 2026. Childcare costs tax credit: €1,737m outturn in 2024 (lower bound), €1,799m forecast for 2025 and 2026 (centre and upper bound), instead of €1,721m (PLF 2025). Crèche places: €695m for 35,000 places, i.e. about €19,900 of public operating cost per place per year (Cour des comptes, 2024). Places of all types: €1.5bn for 100,000 places (35% in crèches, 65% with registered childminders), i.e. €15,000 per place (central and high); average public cost of an existing formal childcare place: €13,000 (low). Investment (€30,000 to €50,000 per crèche place, according to the Cour des comptes) is one-off and not counted. The Cour des comptes notes that the shortage of professionals makes these creations uncertain. Convention: the cost of a place is the gross public cost of a new place in a full year, without deducting what the childcare arrangements it replaces used to cost. The Institut Montaigne (2022) counts crèche places net of this substitution, from −€2,500 (substitution for home-based childcare) to +€5,000 per place per year (substitution for registered childminders), figures it attributes to Insee according to Le Parisien: a third-hand benchmark, with a different scope, not counted. A place that meets unmet demand replaces no subsidised arrangement.

Old age and long-term carecostable, spending
  • New nursing home (Ehpad) places, in operation. For 10,000 more nursing home (Ehpad) places in operation (health insurance care and APA in establishments, excluding accommodation and investment): a cost of €0.27bn per year (0.22 to 0.27) (coefficient calculated by the editors from PLFSS 2026, annexe 6 « Situation financière des établissements de santé et des établissements médico-sociaux », tableau « Evolution des ressources issues de la branche autonomie par type d'établissements et services » (personnes âgées), pages 37-38). Current value: 0 places. Range measured by the source: from 0 places; beyond it, linear extrapolation is flagged. Coefficient calculated by the editors, proportional to the number of places: health insurance care funding for care homes (EHPAD) (€13.36bn in 2024) ÷ installed places (609,970 at end-2023), about €21,900 per place, plus the APA (personalised autonomy allowance) in establishments (€4,940 per beneficiary per year in 2023). Lower bound: care only. No upper bound.
  • Full Libault package (new measures + demographics), additional need in 2030: a cost of €9.2bn per year.
  • Libault package, 2024 stage: a cost of €6.2bn per year.
  • Care home renovation plan (€3bn over 10 years): a cost of €0.3bn per year.
  • Effort for home help services: a cost of €0.55bn per year.
Current law, horizon, limitations and 6 sources

Current law. Autonomy branch of social security, personalised autonomy allowance (APA), three-part pricing of care homes (EHPAD) (care, dependency, accommodation). Tax credit for employing a home worker: €6,856m forecast for 2025 (CGI (General Tax Code), Art. 199 sexdecies), partly for old age.

Horizon. €bn per year in 2018 GDP-point equivalent, compared with 2018; not restated for the 2032 horizon

Limitations. Parameter “Ehpad places” (18 October 2026): annual public operating cost of a place, calculation by the editors: €13,359m of recurrent care funding from the autonomy branch to nursing homes (Ehpad) in 2024 (PLFSS 2026, annex 6) ÷ 609,970 places installed at the end of 2023 (Drees), i.e. about €21,900 (€0.22bn for 10,000 places, lower bound), plus the APA (personalised autonomy allowance) in establishments, €4,940 per beneficiary in 2023 (Drees; €0.05bn for 10,000 places), hence €0.27bn. Relating 2024 funding to 2023 places slightly overstates the cost per place; conversely, residents' community healthcare, non-recurrent funding and the increase in allocations since 2024 are not counted. Not counted: accommodation (paid by residents or by departmental social assistance), housing benefits and investment (construction, aides à la pierre), which is one-off. Residents of new places could have been helped at home (APA at home): this substitution is not deducted (static costing). The Libault report dates from 2019: part of it has been implemented since (Ségur, creation of the autonomy branch in 2021), so the extra cost compared with current policies is below €9.2bn. The figure includes demographics, to be removed if the reference scenario already incorporates them. The 25% increase in the staffing ratio in care homes (EHPAD) (13 FTE per 100 residents) is not isolated in euros in the passages read. No official costing later than 2023 has been read. Annual public cost of an additional place in a care home (Ehpad) (health insurance care package, dependency) not found in a recent official source: commitments expressed in numbers of places remain not costed.

Pensions

Age and contribution period

Contribution periodcostable, spending
  • Required contribution period for a full-rate pension. For 1 quarter less of required contribution period (compared with 172): a cost of €0.75bn per year (0.5 to 1) (coefficient calculated by the editors from Service-public.fr, fiche F35063 (durée d'assurance par génération)). Above the current value, for 1 quarter more of required contribution period (compared with 172): a gain of €1.3bn per year (1 to 1.3); coefficient specific to lengthening: the Cour des comptes costs four more quarters at €5.2bn for the pension system in 2035, i.e. €1.3bn per quarter (compared with €3.9bn for four fewer quarters), and one more quarter reaches €1bn of effect as early as 2030, compared with 2034 for one fewer quarter (lower bound: €1bn); editors' calculation, 2035 horizon. Current value: 172 quarters. Range measured by the source: 168 to 176 quarters; beyond it, linear extrapolation is flagged. Range measured by the Cour des comptes: from four quarters fewer (168) to four quarters more (176), each direction with its own coefficient. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 6 sources

Current law. Contribution period required for a full-rate pension (Touraine Act of 2014, accelerated in 2023, adjusted by the LFSS 2026): 170 quarters for 1964 and for those born from January to March 1965; 171 for those born from April to December 1965; 172 quarters (43 years) for 1966 and later. Age at which the reduction (décote) no longer applies: 67.

Horizon. Full year 2032, €bn per year, balance of the pension system, static. The effect keeps growing after 2032.

Limitations. (1) The phase-in is slow. According to the Cour des comptes, one quarter less only reaches €1bn of impact on the balance in 2034. Four quarters less cost €3.9bn in 2035 and €7.7bn in 2045. Hence a value of €0.5bn to €1bn per quarter in 2032. (2) The figure of €9bn for 160 quarters is a simple linear extrapolation, to be handled with caution: the impact assessment of the 2014 law estimated at €10.4bn per year in 2040 (€5.4bn in 2030), in 2011 euros, the saving from moving from 166 to 172 quarters for all schemes, including supplementary schemes; for the general scheme and the non-balanced basic schemes, €5.6bn in 2040. (3) Macroeconomic effect, presented separately: the Cour des comptes counts €3.2bn of other lost revenue in 2035 for 4 quarters less. (4) Declared overlaps with the legal retirement age (the age coefficient partly includes the effect of the contribution period, through the COR costing; a joint increase of both partly delays the same retirements, common share measured according to the Cnav, report annexed to the PLFRSS 2023, table 5) and with long careers. (5) Long careers react asymmetrically: 80% postpone their retirement if the period increases, 40% bring it forward if it decreases (Prisme model). The figure of €9bn for 160 quarters is a linear extrapolation: it is not used as an option. (6) The « 168 trimestres, 3,9 Md€ » [“168 quarters, €3.9bn”] option is withdrawn: it is a 2035 figure, which gave the same proposal a different value from the parameter (€3bn in 2032). (7) Lengthening. The Cour des comptes costs a one-year lengthening (176 quarters for the 1969 cohort) at €5.2bn for the pension system in 2035, and €9.7bn for public finances as a whole. The increase is therefore costed at its own coefficient, €1.3bn per quarter (5.2 / 4), and not at the coefficient of the decrease: one more quarter reaches €1bn of effect as early as 2030, one quarter less only in 2034 (annex no. 3), hence the lower bound of €1bn. This coefficient is for 2035, not 2032: it slightly overstates the 2032 gain. No costing read beyond 176 quarters. No costing read either for the age at which the reduction (décote) is cancelled (67).

Long careerscostable, spending
  • Broadening of the 2023 type (strengthening, then adjustments voted by Parliament) (2030 horizon): a cost of €1.3bn per year (0.6 to 1.3).
Current law, horizon, limitations and 4 sources

Current law. Early retirement for long careers (Art. L. 351-1-1 CSS (Social Security Code), widened in 2023): retirement possible from 58 at the earliest (thresholds at 58, 60, 62 and 63 depending on the age at which the career began), with the required period and 4 to 5 quarters contributed before 16, 18, 20 or 21. The LFSS 2026 adjusts only the 62 threshold and the required period for the 1964 to 1968 cohorts (COR 2026).

Horizon. Full year 2030 for the 2023 broadening (the only official costing available). The 2016 total is a size benchmark, not the cost of a measure.

Limitations. No recent official costing (2024-2026) has been found for a standard broadening, for example retirement at 60 for all those who started before 20. The cost depends on the age thresholds and the required period. It interacts strongly with the legal retirement age: the Cour des comptes assumes the thresholds unchanged when the age falls and shifted when it rises, which explains the asymmetry of its costings. The figure of €6.1bn (2.3% of total pensions in 2016) predates 2023 and serves only to give an order of magnitude. A current value would require a DSS source or the old-age PQE of the PLFSS 2027. Total early-retirement pensions (€6.1bn in 2016) are a size benchmark, not the cost of a measure.

Arduous workcostable, spending
  • Reinstate the 4 factors removed in 2017 (return to the 2014 scope), 2030 horizon: a cost of €1.7bn per year (0.3 to 2).
Current law, horizon, limitations and 5 sources

Current law. Occupational prevention account (C2P), 6 factors since the 2017 ordinances, which removed manual handling of loads, arduous postures, vibrations and chemical agents. The points allow retirement to be brought forward by up to 2 years, hence from 62 since the 2023 Act. For ergonomic factors, the 2023 Act created a fund for investment in the prevention of occupational wear and tear (FIPU), financed by the occupational injuries and diseases branch (AT-MP).

Horizon. 2030 horizon, €bn per year. No recent costing exists for 2032.

Limitations. Low confidence: the sources are 9 to 12 years old. The impact assessment of the 2014 law gave €2bn in 2030 (€2.5bn in 2040) for the entire 10-factor account (18.2% of workers exposed): as an upper bound for only the 4 factors removed in 2017, it overstates. Rexecode gave €1.7bn for a doubling of beneficiaries (figure read in the Institut Montaigne page, the Rexecode study has not been found). The lower bound, about €0.3bn in 2050 (2017 euros), is the costing cited by Montaigne for a broadening of the scope. The cost depends mainly on the exposure thresholds and the use of the points: early retirement, training or part-time work. The share funded by the AT-MP (occupational accidents and diseases) branch is not pension spending. No figure read gives the current amount of the C2P or the FIPU.

Pension amounts

Minimum pensioncostable, spending
  • Increase in the guaranteed minimum pension for a full career. For +€100 a month in the guaranteed minimum pension for a full career: a cost of €3.5bn per year (1.8 to 5.2) (coefficient calculated by the editors from Cour des comptes, Situation financière et perspectives du système de retraites, note 14). Current value: 0 €/month. Range measured by the source: 0 to 100 €/month; beyond it, linear extrapolation is flagged. Range calculated for +€100 a month. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 7 sources

Current law. Contributory minimum (MiCo) of the aligned schemes, Art. L. 351-10 CSS (Social Security Code): €733.03 per month in 2024 (Cour des comptes), pro-rated if the career is incomplete. The 2023 Act raised the MiCo by €25 and the enhanced MiCo by €75, with a target of a total pension of 85% of the net SMIC (€1,193 in September 2023) for a full career contributed at the SMIC. The enhanced MiCo is now indexed to the SMIC. Separately, the minimum old-age income (ASPA) is €12,523.14 per year for a single person (service-public, February 2026).

Horizon. Full year, €bn per year. The lower bound is costed for 2030, the upper bound in 2018-2022 euros.

Limitations. High uncertainty. The lower bound corresponds to the 2023 measure: up to +€100 for new retirees and an uprating of the minimum amounts for current retirees, €1.8bn in 2030. But only 125,000 retirees actually receive the €100. The upper bound applies Montaigne's rule to the MiCo (minimum contributory pension): €7.0bn of spending × 100 / €134 average MiCo. It overstates the cost, as it assumes that all recipients receive the increase. The central value is the midpoint of this range, not a published figure. No published costing covers +€100 for everyone: the Institut Montaigne (€1.3bn in 2022) costs a €20-per-month increase in the minimum (its rule gives €6.5bn for +€100 with an average MiCo of €110); the Senate (€1.8bn in 2030) costs the 2023 measure, on average €400 a year for 1.8 million retirees; iFRAP (€2bn) publishes no method. Options withdrawn on 13 October 2026, because they were costed against the law as it stood in 2022: a minimum pension of €1,400 for a full career (Montaigne, €12.05bn, on the MiCo and the enhanced MiCo prior to the 2023 act) and the minimum old-age income raised to about €1,041 to €1,100 (Montaigne, €0.67bn, on an ASPA of €917 per month). The 2026 law has changed both starting points (enhanced MiCo at €903.93 per month on 1 January 2026, ASPA at €1,043.59) and no source read costs these measures against it: the positions targeting them are not costed. The target of 85% of the net SMIC in the 2023 act is an objective monitored in a report to Parliament (total pension, basic and supplementary, for a full career at the SMIC), not a guaranteed amount. The cost also depends on the population covered (existing or new retirees), on the condition of 120 contributed quarters and on the inclusion of civil servants (guaranteed minimum, €1.6bn in 2020) and farmers. External benchmark, not counted: the Institut Montaigne (2024 legislative elections) costs raising the minimum contributory pension to a SMIC of €1,600 net for full careers at €13.9bn by 2027 (+€286 per month applied to €7.7bn of minimum contributory pension spending, general scheme). Positions costed on this parameter use its central value: the upper bound overstates the cost.

Pension indexationcostable, spending
  • Uprating of basic pensions above (or below) the current rule, cumulative points up to 2032. For 1 point of uprating of all basic pensions above inflation, cumulated over the upratings from 2028 to 2032 (permanent effect on the level of pensions); a one-point gap for one year costs the same: a cost of €2.35bn per year (2.08 to 2.98) (coefficient calculated by the editors from Commission des comptes de la Sécurité sociale, Les comptes de la Sécurité sociale, résultats 2025, prévisions 2026 et 2027 (octobre 2026), fiche 2.5, tableau 1). Current value: 0 point. Range measured by the source: -10 to 10 point; beyond it, linear extrapolation is flagged. Reading key: a one-point gap from the current rule for one year is worth about €2.35bn net (2.1 to 3.0; 1% of the €297.9bn of basic pensions in 2025). A rule maintained from 2028 to 2032 accumulates five upratings: common inflation assumption of 1.75% per year (1.5 to 2.0%), wage-price gap of 0.7 points per year (reference scenario of the COR (pensions advisory council), June 2026). Coefficient proportional to the basic pension bill (editors' calculation): range of ±10 cumulative points, beyond that linear extrapolation, flagged as such.
  • Uprating of basic pensions above €2,000 a month only, gap with the current rule in cumulative points up to 2032. For 1 additional point of uprating (or 1 point less, a saving) for basic pensions above €2,000 per month only, cumulated from 2028 to 2032; a one-year freeze is worth −1.75 points (inflation under the common assumption): a cost of €1.27bn per year (1.12 to 1.61) (coefficient calculated by the editors from Commission des comptes de la Sécurité sociale, Les comptes de la Sécurité sociale, résultats 2025, prévisions 2026 et 2027 (octobre 2026), fiche 2.5, tableau 1). Current value: 0 point. Range measured by the source: -10 to 0 point; beyond it, linear extrapolation is flagged. Reading key: one point of uprating missed for one year by pensions above €2,000 brings in about €1.27bn net (1.1 to 1.6); a one-year freeze, 1.75 points of inflation (common assumption, 1.5 to 2.0%), about €2.2bn. Editors' calculation based on the IPP (€2.8bn net for a freeze at 2.5% inflation). Range: from −10 to 0 cumulative points (freezes and under-indexation); beyond that, linear extrapolation, flagged as such.
  • Uprating of basic pensions above the median only, gap with the current rule in cumulative points up to 2032. For 1 additional point of uprating (or 1 point less, a saving) for basic pensions above the median only, cumulated from 2028 to 2032; a one-year freeze is worth −1.75 points (inflation under the common assumption): a cost of €1.67bn per year (1.48 to 2.12) (coefficient calculated by the editors from Commission des comptes de la Sécurité sociale, Les comptes de la Sécurité sociale, résultats 2025, prévisions 2026 et 2027 (octobre 2026), fiche 2.5, tableau 1). Current value: 0 point. Range measured by the source: -10 to 0 point; beyond it, linear extrapolation is flagged. Reading key: one point of uprating missed for one year by pensions above the median brings in about €1.67bn net (1.5 to 2.1); a one-year freeze, 1.75 points of inflation (common assumption, 1.5 to 2.0%), about €2.9bn. Editors' calculation based on the IPP (€3.7bn net for a freeze at 2.5% inflation). Range: from −10 to 0 cumulative points (freezes and under-indexation); beyond that, linear extrapolation, flagged as such.
  • Sustainability coefficient of −0.4 point applied to basic pensions from 2026 to 2029 (average annual gain): a gain of €3.1bn per year.
Current law, horizon, limitations and 11 sources

Current law. Basic pensions are uprated on 1 January each year in line with observed inflation (annual average of prices excluding tobacco, Art. L. 161-25 CSS (Social Security Code)), and cannot fall in nominal terms. Uprating of +0.9% on 1 January 2026: the freeze planned in the PLFSS 2026 was not adopted; +1.7% forecast for 1 January 2027 (Commission des comptes de la Sécurité sociale (Social Security Accounts Committee), October 2026). Statutory benefits of the basic schemes: €297.9bn in 2025 (288.0 in 2024). Supplementary pensions (Agirc-Arrco) are managed by the social partners, with a sustainability coefficient of 0.4 points below inflation in the 2023-2027 agreement. The PLF and PLFSS 2027 under discussion (October 2026) mention €6bn of savings on basic pensions: this is not yet current law.

Horizon. Full year 2032, €bn per year in 2025 euros, compared with the current rule (uprating in line with inflation). Assumption common to all indexation rules (pensions, public sector pay index point, grants; decision of 18 October 2026): inflation of 1.75% per year (1.5 to 2.0%), uprating of 1.7% on 1 January 2027, first effect of a measure of the five-year term on 1 January 2028; the 2032 cost accumulates the five upratings from 2028 to 2032.

Limitations. (1) Coefficient per point (calculation by the editors). Upper bound, gross: 1% of the statutory benefits of the basic schemes in 2025, €297.9bn (Commission des comptes de la Sécurité sociale, October 2026), i.e. €2.98bn; the €288bn « prévus pour 2025 » [“forecast for 2025”] by the Cour des comptes (Court of Audit) (€2.9bn per point) correspond to the actual level of 2024. Central value, net: the same total × the IPP's net/gross ratio for a freeze (5.2 / 6.6 = 0.79, once income tax, CSG and means-tested benefits are deducted), i.e. €2.35bn. Lower bound: the IPP freeze related to one point on its own scope, existing retirees only (5.2 / 2.5 = €2.08bn net). Under-indexation reads as a saving, with the same unit value. (2) Freeze recalculated. The IPP costs the 2027 freeze with 2.5% inflation (€6.6bn gross, 5.2 net); with the 1.7% uprating forecast for 2027 (CCSS, October 2026), the same freeze would be worth €4.5bn gross and 3.5 net (proportional calculation), and about €4.1bn net for a one-year freeze at 1.75% inflation, the common assumption. The options “full freeze in 2027” (€5.2bn) and “freeze above €2,000” (€2.8bn), costed at 2.5% inflation for a single year, were replaced on 18 October 2026 by parameters in cumulative points: a freeze of all pensions is costed on the general parameter (−1.75 points per year of freeze), a freeze above a threshold on the threshold parameter (−1.75 points per year of freeze). (3) Thresholds (IPP, note no. 129): restricting the freeze to pensions above the median reduces the net saving from €5.2bn to €3.7bn, at €2,000 to €2.8bn, at €3,000 to €1.2bn; hence, per point of inflation, €1.48bn (median) and €1.12bn (€2,000) on the IPP's scope (lower bounds), scaled to the 2025 total like the general parameter (central values), gross as upper bound (same ratio 6.6 / 5.2, which slightly underestimates the gross figure for high pensions, which are more heavily taxed). (4) Cumulation. A rule maintained from 2028 to 2032 is costed by the sum of the annual gaps, in points, applied to the 2025 total: compound interest (less than 0.1 point over five years) and the turnover of retirees are disregarded (pensions awarded after 2028 undergo only part of the gaps, which slightly overstates the amount; the lower bound, on the scope of existing retirees only, partly corrects this). The €2,000 threshold is assumed fixed in euros. (5) Indexation on wages: the annual gap used is that of real earned income in the reference scenario of the COR (pensions advisory council), 0.7 point per year until 2038 (annual report of June 2026), i.e. 3.5 cumulative points in 2032 and about €8.2bn net (7.3 to 10.4). The Institut Montaigne (2024 legislative elections) used 1.1 points (2001-2023 average): €3.4bn gross per uprating, 2.5 net, same order of magnitude per point. The option “indexation on wages” (a single uprating at 1.1 points) was withdrawn on 18 October 2026 in favour of the parameter. (6) Supplementary pensions (€401bn of basic and supplementary benefits in 2025) are not set by law. (7) Option “sustainability” (Cour des comptes): costed by the Cour des comptes as a 2026-2029 average on its own inflation trajectory, not recalculated.

Survivors' pensionscostable, spending
  • Change in the total of survivors' pensions. For 1% change in the total of survivors' pensions: a cost of €0.39bn per year (coefficient calculated by the editors from COR, Rapport annuel juin 2026, chap. 2). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 5 sources

Current law. Rules differ between schemes. General scheme: 54% of the deceased's basic pension, from age 55, subject to a means ceiling (€25,001.60 per year for a single person, €40,002.56 for a couple, service-public page checked on 1 January 2026). Civil service: 50% of the basic pension. Agirc-Arrco: its own rules, not read here. In 2025, €39bn of derived-right pensions (1.3% of GDP) for 4.4 million beneficiaries (2022), 88% of them women.

Horizon. €bn per year, 2025 base. For 2032, to be uprated like total pensions (stable as a share of GDP according to the COR (pensions advisory council)).

Limitations. No official costing read for a specific option: harmonisation across schemes, abolition of the means-test ceiling in the general scheme, single rate. The unit cost schedule, 1% of the total, serves only to translate an option once its scope has been estimated. Harmonisation can be designed at zero cost, with winners and losers. The effects are concentrated on women: survivor's pensions account for 19.3% of their average total pension in 2024 (COR).

Taxation of pensioners

10% allowance on pensionscostable, revenue
  • Abolition of the 10% allowance on pensions: a gain of €5bn per year (4.8 to 5.4).
  • Ceiling lowered to €2,000 per household: a gain of €2.15bn per year.
  • Ceiling lowered to €2,500 per household (€2,545 according to the IPP): a gain of €1.51bn per year (1.5 to 1.51).
  • Replacement by a flat allowance of €2,000 per pensioner (PLF 2026, art. 6, excluding the tightening of art. 157 bis): a gain of €0.88bn per year (0.5 to 1.1).
Current law, horizon, limitations and 7 sources

Current law. 10% allowance on pensions and retirement income for income tax (Art. 158, 5-a CGI (General Tax Code), created by the Finance Act for 1978). Floor of €454 per pensioner and ceiling of €4,439 per household on 2025 income. The PLF 2026 (Art. 6) sought to replace it with a flat €2,000 per pensioner: the measure was rejected and the allowance maintained. The question returns in the 2027 budget, under discussion in October 2026.

Horizon. Full year 2024 (outturn, lower bound of the abolition) to 2027, €bn per year of revenue. Not uprated for 2032 (it would follow pensions).

Limitations. Lower bound: the gross tax cost outturn in 2024, €4.80bn (Voies et moyens, volume II, PLF 2026, tax expenditure 120401; €5.32bn forecast for 2025), instead of the €4.67bn forecast for 2026, a forecast that appears to incorporate the €2,000 flat allowance of the PLF 2026, which was rejected (the same volume records −€551m for this measure in 2026; under unchanged law, about €5.2bn, a calculation not verified in the document). The PLF 2027 (art. 3) proposes a sub-ceiling of €3,000 for retirement pensions (€1.4bn in 2027, €1.1bn from 2029, impact assessment): no position takes it up, no option. The IPP costs abolition at €5.4bn net in 2027: €5.0bn of income tax plus €0.4bn less in housing benefits. This second effect penalises low-income retirees and calls for mitigating measures. For the €2,000 flat allowance, the government announced €1.5bn in 2027, of which €0.4bn from art. 157 bis, i.e. €1.1bn for the flat allowance itself. The IPP estimates it at €875m in steady state and at slightly over €0.5bn in reality. The €2,000 and €2,500 ceilings are costed by the DLF for the Senate (€2.148bn and €1.514bn).

  • PLF 2026, Évaluation des voies et moyens, tome II (dépenses fiscales), dépense fiscale 120401, 2025-10: « 120401 – Abattement de 10 % sur le montant des pensions (y compris les pensions alimentaires) et des retraites ; Impact budgétaire (en millions d'euros) Réalisation 2024 : 4 803 ; Prévision 2025 : 5 322 ; Prévision 2026 : 4 665 ; Bénéficiaires 15 085 237 Ménages »
  • Sénat, Rapport général PLF 2026, tome II, article 6, 2025-11: « Selon le tome II du rapport sur les voies et moyens, en 2026, l'abattement de 10 % sur les pensions et retraites constitue la troisième dépense fiscale la plus coûteuse (4,67 milliards d'euros) »
  • Sénat, Rapport général PLF 2026, tome II, article 6 (données DLF), 2025-11: « un abaissement du plafond à 2 500 euros (contre 4 399 euros en 2025) par foyer fiscal générerait, selon les données transmises au rapporteur général par la direction de la législation fiscale, un gain de 1,514 milliard d'euros en régime de croisière [tableau : plafond à 2 000 euros, total 2 148] »
  • Sénat, Rapport général PLF 2026, tome II, article 6 (évaluation préalable et IPP), 2025-11: « « estimé à 1,2 milliard d'euros en 2026 et à 1,5 milliard d'euros en 2027, dont 0,4 milliard d'euros au titre du resserrement du dispositif prévu à l'article 157 bis du CGI » […] l'Institut des politiques publiques (IPP) estime le surcroît de recettes fiscales obtenu à 875 millions d'euros en régime de croisière »
  • IPP, Note n° 129, 2026-09: « La suppression de l’abattement de 10 % (A1) améliore le solde des finances publiques de 5,4 milliards d’euros, dont 5,0 milliards de recettes supplémentaires d’impôt sur le revenu et 0,4 milliard de moindres dépenses de prestations sociales (aides au logement). »
  • IPP, Note n° 129, scénario A2, 2026-09: « son plafond annuel par foyer fiscal est ramené à 2545 euros (contre 4439 euros pour les revenus de 2025), montant calibré sur le rendement de C2 [1,5 Mde] »
  • IPP, Note n° 129, « Comment mettre à contribution les retraités pour le redressement des finances publiques ? » (page 3 du PDF), 2026-09: « Institué par la loi de finances pour 1978, il s’applique automatiquement au montant des pensions de retraite, de réversion et d’invalidité pour le calcul de l’impôt sur le revenu (art. 158, 5-a du CGI). Pour les revenus 2025 déclarés en 2026, l’abattement est plafonné à 4439 euros par foyer fiscal, avec un plancher de 454 euros par pensionné ; ces limites sont revalorisées chaque année comme la première tranche du barème. »
CSG (general social contribution) on pensionerscostable, revenue
  • Median rate (6.6%) and full rate (8.3%) raised to 9.2% (net): a gain of €4.2bn per year.
  • Full rate alone raised from 8.3% to 9.2% (net): a gain of €1.5bn per year.
Current law, horizon, limitations and 5 sources

Current law. CSG on pensions (Art. L. 136-8 CSS (Social Security Code)) has four rates depending on reference taxable income: 0%, 3.8% (reduced rate), 6.6% (median rate) and 8.3% (full rate). 2026 thresholds for one share (2024 RFR): €13,048, €17,057 and €26,472 (€12,817, €16,755 and €26,004 in 2025). Earned income bears 9.2%. In addition there are the CRDS (0.5%, except for those exempt) and the CASA (0.3%, at the median and full rates). Yield of the CSG on retirement pensions: €25.4bn in 2025, €26.5bn forecast for 2026 (CCSS, October 2026).

Horizon. Year 2027 (IPP, TAXIPP, static with second-round effects), €bn per year. Not uprated for 2032 (it would follow total pensions).

Limitations. Net yields: aligning the two rates loses about €0.6bn through a second-round effect, because the increase falls on the deductible portion and reduces income tax. A cut in retirees' CSG, for example a return of the full rate to 6.6%, has no costing read. As an indication, the full rate yields about €1.7bn net per point (1.5 / 0.9), derived from scenario C2, but not beyond the simulated range. The burden falls mainly on the middle classes (median rate, +2.6 points). Revenue from CSG on retirement pensions (basic and supplementary schemes): €25.4bn in 2025, €26.5bn forecast for 2026 (CCSS, October 2026, table 3); the value of one point on all replacement income (€3.90bn in 2025, of which about 90% pensions) is under the CSG lever. Thresholds updated on 18 October 2026 (service-public.fr, fact sheet F2971); the IPP costings remain those of its note, on 2025 legislation.

  • Service-public.fr, fiche F2971, CSG et CRDS sur les revenus d'activité et de remplacement (vérifiée le 1er juin 2026), 2026-06-01: « Taux zéro : jusqu'à 13 048 € ; Taux réduit (3,8 %) : de 13 049 € à 17 057 € ; Taux médian (6,6 %) : de 17 058 € à 26 472 € ; Taux normal (8,3 %) : plus de 26 472 € »
  • Commission des comptes de la sécurité sociale, octobre 2026, fiche 1.3, tableau 3, 2026-10: « Retraites : 2024 24 812 (4,7 %) ; 2025 25 439 (2,5 %) ; 2026 (p) 26 504 (4,2 %) ; 2027 (p) 27 558 (4,0 %). « La CSG sur les revenus de remplacement serait assise en 2026 à 90,2% sur les pensions versées par les régimes d'assurance vieillesse (de base et complémentaires), à 7,1% sur les pensions d'invalidité et les indemnités journalières, ainsi qu'à 1,9% sur les indemnités d'assurance chômage. » »
  • IPP, Note n° 129, encadré droit en vigueur, 2026-09: « Les revenus d’activité supportent une CSG de 9,2 % (dont 6,8 points déductibles du revenu imposable). Les pensions relèvent de quatre taux – 0 %, 3,8 %, 6,6 % et 8,3 % – selon le RFR du ménage et son nombre de parts (art. L. 136-8 du CSS) ; pour 2025, les seuils pour une part sont de 12 817, 16 755 et 26 004 euros. »
  • IPP, Note n° 129, rendements, 2026-09: « porter à 9,2 % les taux de 6,6 % et 8,3 % rapporterait 4,2 Mde. Là encore, un effet de second tour réduit le produit de la mesure d’environ 0,6 Mde »
  • IPP, Note n° 129, scénario C2, 2026-09: « Nous choisissons ainsi un scénario où seul le taux plein de CSG des retraités est aligné sur celui des actifs (C2) : il porte ce taux de 8,3 % à 9,2 % pour une recette nette de 1,5 Mde. »

Organisation of the system

Special pension schemes, points-based systemnot costable
Current law, horizon, limitations and 3 sources

Current law. Since 1 September 2023 (Act of 14 April 2023, Art. 1), new recruits at the RATP, the IEG, the Banque de France, the CESE, and notaries' clerks and employees are affiliated to the general scheme: these special pension schemes are being phased out. Existing members remain in them. Other special pension schemes remain (SNCF closed since 2020, seafarers, mines, State workers…). The State contributes just under €8bn to balancing 17 special pension schemes and paid €45bn in 2023 to its civil servants' scheme. Agirc-Arrco already works on a points basis, but there is no universal points-based system (2019-2020 project abandoned).

Horizon. Very long-term effects, virtually nil in 2032 on a static basis.

Limitations. Closing a scheme to new entrants yields almost nothing for decades. Montaigne 2022 costed the abolition of the special pension schemes at €5.3bn « à très long terme » [“in the very long term”] (2017 euros), with no effect in 2027. A points-based system is neutral by design: its cost depends on the parameters (value of the point, pivot age, solidarity provisions). It is classified as a rule change, costed only if the parameters are given. Moving civil servants into the general scheme raises the question of the accounting convention for the State scheme (« subvention d'équilibre », balancing subsidy), with no change to the general government balance.

  • Sénat, L'essentiel PLFRSS 2023, article 1er, 2023-03-13: « L’article 1er propose la mise en extinction des principaux régimes spéciaux de retraites […] les personnes recrutées à partir du 1er septembre 2023 par les entreprises et institutions concernées seraient affiliées au régime général. À l’inverse, les personnes déjà affiliées à ces régimes spéciaux y seraient maintenues. »
  • Cour des comptes, Situation financière et perspectives du système de retraites, 2025-02-20: « il participe à l’équilibre financier de 17 régimes spéciaux pour un peu moins de 8 Md€. D’autre part, il finance le régime de ses fonctionnaires civils et militaires au moyen d’une contribution de 45 Md€ en 2023. »
  • Institut Montaigne 2022, fiche Macron 65 ans (archive), 2022-04: « La suppression des régimes spéciaux permettra de générer, à très long terme, de l’ordre de 5,3 Md€ d’économies supplémentaires (en € de 2017). Ces 5,3 Md€ ne sont donc pas pris en compte dans ce chiffrage. »
Funded sharecostable, spending
  • Basic pension contribution points shifted to funded pensions. For 1 point of basic pension contributions diverted from pay-as-you-go to a funded account (transition cost): a cost of €6.2bn per year (4.8 to 7.6). Current value: 0 point. Range measured by the source: 0 to 1 point; beyond it, linear extrapolation is flagged. The DSS costs the yield of one contribution point; beyond one point, linear extrapolation, flagged as such.
  • Public endowment paid into a retirement account at each birth. For €1,000 paid at each birth (643,905 births in France in 2025): a cost of €0.64bn per year (coefficient calculated by the editors from Céreq Bref n° 487-488, « S'insérer dans un monde instable : le diplôme reste une valeur sûre » (enquête 2024 auprès de la Génération 2021)). Current value: 0 € per birth. Range measured by the source: from 0 € per birth; beyond it, linear extrapolation is flagged. Exact product of the number of births (Insee, 2025) and the amount paid: no upper bound.
  • Public payment into a retirement account at the first job. For €1,000 paid to each young person at their first job (773,000 leavers from initial education per year, Céreq): a cost of €0.77bn per year (coefficient calculated by the editors from Céreq Bref n° 487-488, « S'insérer dans un monde instable : le diplôme reste une valeur sûre » (enquête 2024 auprès de la Génération 2021)). Current value: 0 € par jeune. Range measured by the source: from 0 € par jeune; beyond it, linear extrapolation is flagged. Product of the annual number of first-time leavers from the education system (Céreq, Génération 2021: 773,000) by the amount paid: no upper bound.
Current law, horizon, limitations and 5 sources

Current law. The mandatory system is pay-as-you-go. The only funded exceptions: the RAFP (civil service, €43.3bn in provisions) and the FRR (€21.2bn in assets at end-2023, which pays €1.45bn per year to the Cades in 2025). Voluntary retirement savings (PER) accounted for €8.3bn in benefits and €18.5bn in contributions in 2022, for outstanding assets of €292.7bn at end-2023. Their tax and social advantages cost about €1.8bn per year.

Horizon. €bn per year, 2025 base: revenue loss for the pay-as-you-go scheme during the transition, pensions unchanged.

Limitations. Parameter “payment at the first job” (18 October 2026): arithmetic coefficient, 773,000 young people who left the education system for the first time in 2021 (Céreq, Génération survey) × amount paid, i.e. €0.77bn per year for €1,000; each leaver is assumed to obtain a job one day and to receive the payment once, the annual flow of first jobs then being equal to that of leavers. Leavers in 2017 numbered 746,000 (Céreq). Any switch to funded pensions has a double cost: the diverted contributions no longer finance current pensions, which the State must continue to pay. The range uses the yield of one point of contribution to the general scheme: €4.8bn (employee share, uncapped), €6.2bn (below the ceiling) and €7.6bn (employer share, uncapped), according to the DSS. A public endowment to a sovereign pension fund is costed as a lump sum equal to the amount paid in, hence the value 0 to be filled in according to the programme. Future financial returns are not counted in a static costing. A mere new tax incentive for the PER (retirement savings plan) is costed as a tax expenditure, from the €1.8bn benchmark. A public endowment to a funded pension fund is costed at the amount paid in, which is announced by the candidate; the cost of retirement savings incentives (€1.8bn) is a benchmark. Parameter “endowment at birth”: arithmetic coefficient, number of births in France in 2025 (Insee, 643,905) × amount paid, i.e. €0.64bn per year for €1,000. The number of births has fallen every year since 2010: the cost follows. Newborns of French nationality are slightly fewer than births in France. The payment is counted as spending (a transfer to the child); if it remained the property of a public fund, it could be classified as a financial transaction, which weighs on the debt and not on the deficit. Tax-exempt family top-ups and the fund's future return are not costed.

Taxes and public finances

Income tax and CSG

Income tax scalecostable, revenue
  • Proportional change in net income tax. For +1% on all net income tax (1% of €96.8bn forecast for 2026): a gain of €0.97bn per year (0.95 to 1.03) (coefficient calculated by the editors from Fipeco, L'impôt sur le revenu et la CSG). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • Freeze of the tax schedule (inflation of about 1%), Bercy costing deemed « vraisemblable » [“plausible”] by the HCFP, ultimately not adopted: a gain of €1.9bn per year (1.1 to 1.9).
  • Top bracket at 60% (instead of 45%) above €250,000 of taxable income per share, static (editors' calculation on DGFiP data by marginal rate): a gain of €1.3bn per year (1 to 1.6) (coefficient calculated by the editors from DGFiP, Impôt sur le revenu : nombre de foyers et montant de l'impôt net par taux marginal de taxation du barème, onglet 2024 (mise à jour du 7 avril 2026)).
  • Income tax reduced by up to €3,000 (exemption below) during the three years following a birth, static (editors' calculation on the DGFiP deciles of reference taxable income per share): a cost of €1.7bn per year (coefficient calculated by the editors from DGFiP, Impôt sur le revenu : répartition de l'impôt moyen par dixièmes de RFR par part, onglet a2024).
Current law, horizon, limitations and 9 sources

Current law. 2026 scale (2025 income), per share: 0% up to €11,600; 11% from €11,601 to €29,579; 30% from €29,580 to €84,577; 41% from €84,578 to €181,917; 45% above. Brackets uprated by 0.9% by the Finance Act for 2026, promulgated on 19 February 2026; the PLF provided for a freeze. In addition there are the CEHR (3% and 4%) and the CDHR, a minimum tax of 20% of reference taxable income (RFR) above €250,000 (single person) or €500,000 (couple), created in the LFI 2025 and extended by the LFI 2026 « jusqu'à ce que le déficit du budget général soit inférieur à 3 % du produit intérieur brut » [“until the general budget deficit is below 3% of gross domestic product”] (Voies et moyens, PLF 2027, volume I, p. 30, line 1440: €369m in 2025, €790m forecast for 2026, €600m in 2027): its continuation is current law. Yield: €94.9bn of net income tax in 2025, €96.8bn forecast for 2026 and €102.5bn for 2027, including PLF 2027 measures (Voies et moyens, PLF 2027); €104.0bn forecast in the PLF 2026, €99.8bn in the amended état A (January 2026).

Horizon. Full year, in 2026 euros. The base is not brought to 2032 value (of the order of +15 to +20% under a nominal growth assumption, not sourced).

Limitations. No official cost schedule has been published for one point of rate per bracket: the breakdown of revenue by bracket is not public (Fipeco: « elle n'est pas possible pour l'IR » [“it is not possible for income tax”]). Costings by bracket (creation of brackets, marginal rate of 50% or more) require a microsimulation (OpenFisca, IPP). Freezing the tax schedule: €1.9bn according to Bercy and the HCFP (Voies et moyens of the PLF 2026), about €1.2bn according to the OFCE (1.1% uprating avoided) and €1.1bn according to the IPP updated in October 2025 (inflation of 1%; €1.4bn in its first estimate, with 1.3%); the gap stems from the inflation assumptions. Lower bound: the lowest figure, that of the updated IPP estimate. Proportional coefficient calculated by the editors: 1% of €94.9bn (2025, lower bound), of €96.8bn (2026, central value) and of €102.5bn (2027, upper bound), according to the Voies et moyens of the PLF 2027. Net income tax includes the 12.8% flat tax (PFU): declared overlap with the income-tax share of the PFU. CDHR (differential contribution on high incomes): the option “extension of the CDHR” (€1.4bn to €1.7bn, one-year extension in the PLF 2026) was withdrawn on 18 October 2026; the 2026 initial finance act (LFI) extended the contribution until the general budget deficit falls below 3% of GDP (Voies et moyens, PLF 2027): maintaining it is status quo; only a strengthening beyond current law (rates, thresholds) would be costed, on the difference, and no source read costs it. Top bracket at 60% above €250,000 per share (option, calculation by the editors, 18 October 2026): the DGFiP (public finances directorate-general) publishes 61,476 households at the 45% marginal rate (family quotient above €180,294) and €12.54bn of net tax (2024 income), with no breakdown above €250,000. With 1.8 shares per household (the top thousandth of RFR per share counts 1.82, the top hundredth 1.73), the average tax per share (€113,300) gives an average taxable income of about €303,600 per share above €180,294 (tax of €57,865 at that threshold, then 45%); a Pareto distribution with the same mean (coefficient 2.46) places 44.7% of these shares above €250,000, with €170,900 on average beyond it: €8.45bn of income above the threshold, and 15 more points yield €1.27bn (centre, rounded to 1.3); 2.0 shares per household give €1.0bn, 1.6 shares €1.6bn. Net tax is taken after tax reductions and credits (which understates income); no behavioural effect, though it is strong at these income levels; capping of the family quotient and CEHR (exceptional contribution on high incomes) ignored. Exemption or reduction of €3,000 for three years after a birth (option, calculation by the editors): the benefit per household is min(tax, €3,000). The DGFiP publishes the average net tax per share and the average number of shares for each tenth of RFR per share (2024 income): about €290 per household in the 6th tenth, €840 in the 7th, €1,640 in the 8th, €3,320 in the 9th and €16,340 in the 10th, nothing below; capped at €3,000, the average of these ten values is €877 per household, which overstates the average benefit (each household being capped, not the average of its tenth). Assuming that the households concerned are distributed like all households, and counting them at most as the children born in three years (643,905 births in 2025, Insee, i.e. 1.93 million over three years), the cost is at most €1.7bn per year. Households of young parents, who are in work, are more often taxable than retirees or students, but have more shares: the distribution assumption may work in either direction.

CSG (general social contribution)costable, revenue
  • Change in all CSG (general social contribution) rates, all bases. For 1 point increase in all CSG rates (all bases): a gain of €15.75bn per year (15.75 to 18) (coefficient calculated by the editors from Fipeco, L'impôt sur le revenu et la CSG). Current value: 0 point. Range measured by the source: -1 to 1 point; beyond it, linear extrapolation is flagged. The source costs the yield of a one-point increase (Fipeco, based on the Commission des comptes de la sécurité sociale). Beyond that, linear extrapolation, flagged as such.
  • CSG rate on earned income. For 1 point of CSG on earned income only (2025 yield, €111bn at the 9.2% rate, divided by 9.2; central value net of the income tax due on the deductible portion, 6.8 points out of 9.2): a gain of €11bn per year (11 to 12.1) (coefficient calculated by the editors from Fipeco, L'impôt sur le revenu et la CSG). Current value: 9.2%. Range measured by the source: 9.2% only; beyond it, linear extrapolation is flagged. Coefficient calculated by the editors from the yield at the observed rate (9.2%): any other value is a linear extrapolation, flagged as such, which ignores behavioural effects.
  • Change in all CSG rates on replacement income (pensions, unemployment benefits, daily sickness allowances). For 1-point increase in each CSG rate on replacement income (value of the point published by the CCSS: €3.90bn in 2025; central value net of income tax, the increase being assumed deductible): a gain of €3.41bn per year (3.41 to 3.9) (coefficient calculated by the editors from Commission des comptes de la sécurité sociale, octobre 2026, fiche 1.3, tableau 6). Current value: 0 point. Range measured by the source: -1 to 1 point; beyond it, linear extrapolation is flagged. The CCSS publishes the value of one point (yield of each base divided by its headline rate); a change of one point at most, beyond that linear extrapolation, flagged as such. The zero rate is not affected (its base does not count in the value of the point).
  • Abolition of CSG (general social contribution) on earned income (2025 yield: €111bn; central value, net of the income tax due on the deductible portion): a cost of €100.7bn per year (100.7 to 111) (coefficient calculated by the editors from Fipeco, L'impôt sur le revenu et la CSG).
  • +1 point on earned income only (derived calculation: €111bn / 9.2, net of the income tax due on the deductible portion): a gain of €11bn per year (11 to 12.1) (coefficient calculated by the editors from Fipeco, L'impôt sur le revenu et la CSG).
  • +1 point on capital income only (derived calculation: €18bn / 9.2): a gain of €2bn per year (1 to 2) (coefficient calculated by the editors from Fipeco, L'impôt sur le revenu et la CSG).
  • CSG and CRDS fully deductible from taxable income (earned and replacement income): income tax lost, static (editors' calculation): a cost of €5.6bn per year (5.5 to 5.8) (coefficient calculated by the editors from Commission des comptes de la sécurité sociale, octobre 2026, fiche 1.3, tableau 6).
Current law, horizon, limitations and 5 sources

Current law. CSG at 9.2% on earned income (1.75% allowance for expenses). On pensions: 0%, 3.8%, 6.6% or 8.3% depending on the RFR (reference taxable income). On capital income, the LFSS 2026 raises the rate from 9.2% to 10.6% for part of financial savings income; it remains at 9.2% for property income, real estate capital gains and life insurance. 6.8 points are deductible from income tax. In addition there is the CRDS at 0.5%. Yield: €157bn in 2025 (Fipeco), €153.8bn in 2024 (CCSS).

Horizon. Full year, in 2025 euros. The base is not brought to 2032 value (of the order of +15 to +20% under a nominal growth assumption).

Limitations. The figure of €18bn per point comes from Fipeco based on CCSS data (2025 yield): it is a gross yield. A CSG increase falls largely on the deductible portion and reduces income tax: the IPP measures this effect at €0.6bn out of €4.8bn gross for retirees' CSG (12.5%). The IPP measures this 12.5% rate on an increase falling entirely on the deductible portion (note no. 129: rates of 6.6% and 8.3% raised to 9.2%, « avec la même part déductible (6,8 points) » [“with the same deductible portion (6.8 points)”]): the correction applies only to the deductible portion of the levy. Cost schedule rule (central = net): for one point added to all rates, assumed deductible like the 2018 increase, 18 × (1 − 0.125) = €15.75bn net per point, 18 as upper bound; this calculation also applies the correction to the share of the capital base subject to the flat-rate levy, where CSG is not deductible (about €2bn per point, 18 / 9.2): it understates the gain by at most €0.25bn per point. For an average point of CSG on earned income, of which 6.8 points out of 9.2 are deductible: 12.1 × (1 − 0.125 × 6.8 / 9.2) = €11.0bn net (editors' calculation). Calculations by base divide the revenue by 9.2%; for replacement income, taxed at lower rates, this calculation (28 / 9.2 = €3.0bn per point) underestimated the base: the CCSS publishes the value of one point, €3.90bn in 2025 (4.04 forecast for 2026), used since 18 October 2026 (parameter “replacement income”; net at the centre: 3.9 × (1 − 0.125) = €3.41bn, editors' calculation, the increase being assumed deductible like that of 2018). The same table publishes €12.03bn per point on earned income (the cost schedule keeps 111 / 9.2 = 12.1, a negligible difference) and 1.91 on capital. Aligning retirees' CSG with that of people in work also falls within the pensions domain: a microsimulation by RFR (reference taxable income) band would be needed. LFSS 2026 measure on capital: €1.2bn according to the Senate, €0.1bn in 2026 then €1.4bn in 2027 according to the government (press figure). The 1.4-point increase on part of financial savings (LFSS 2026) is already in force. CSG rate on earned income (parameter): coefficient calculated by the editors from the 2025 revenue (€111bn ÷ 9.2 = €12.1bn gross per point, 11.0 net of income tax at the centre), as for the digital services tax; any value other than 9.2% is a linear extrapolation, marked as such. The option “+1 point on earned income only” is worth one step. Abolition of CSG on earned income: costed at its 2025 revenue (€111bn, Fipeco based on the CCSS), like other tax abolitions; this gross revenue goes to the upper bound; at the centre, net of the income tax that the disappearance of the deductible portion would generate, the 12.5% correction applying only to that portion (6.8 points out of 9.2): 111 × (1 − 0.125 × 6.8 / 9.2) = €100.7bn (editors' calculation; 97.1 before 18 October 2026, when the correction applied to all 9.2 points). Static costing: no effect on employment and no feedback from VAT or contributions. CSG and CRDS fully deductible (option, 18 October 2026, editors' calculation): the non-deductible portion is 2.4 points at each rate (9.2%, of which 6.8 deductible, on earned income; 8.3 of which 5.9, 6.6 of which 4.2 and 6.2 of which 3.8 on replacement income; the reduced rate of 3.8% is fully deductible), and the CRDS (0.5%) is not deductible. Additional deductions: earned income, (2.4 + 0.5) × 12.03 = €34.9bn (value of the point published by the CCSS); replacement income, 2.4 × 3.9 × s + 0.5 × 3.9, where s is the share of the base excluding the reduced rate, not published, taken between 0.75 and 1 (€9.0bn to €11.3bn). The lost tax applies to these deductions the 12.5% rate that the IPP measures for the deductible portion of retirees' CSG (the only one read, which the cost schedule also applies to earned income): 4.36 + 1.12 = €5.5bn (lower bound), 5.6 at the centre (s = 0.875), 5.8 (upper bound). Capital income taxed at the flat rate (without deduction) and property income not counted. The average tax rate on deductions on earned income may differ from 12.5%: no source read measures it.

Family quotientcostable, revenue
  • Cap on the benefit per half-share. For €100 reduction in the cap per half-share (derived from IGF-IGAS 2026: €580m for a 10% reduction, i.e. about €181; lower bound derived from the PLF 2014: €1.03bn for −€500, in 2014 euros): a gain of €0.32bn per year (0.21 to 0.32) (coefficient calculated by the editors from IGF et IGAS, Revue de dépenses relative à l'efficacité des politiques familiales, juin 2026 (page 31 du PDF)). Current value: €1,807. Range measured by the source: €1,307 to €1,807; beyond it, linear extrapolation is flagged. Range measured: 10% reduction in the €1,807 cap (IGF-IGAS, 2026), and €500 reduction (PLF 2014, lower bound). For an increase or beyond −€500, linear extrapolation, flagged as such.
  • Raising of the cap on the benefit per half-share. For +€100 on the cap per half-share (derived from the CPO 2024: €0.8bn from €1,759 to €2,000; €2.1bn from €1,759 to €2,500): a cost of €0.28bn per year (0.28 to 0.33) (coefficient calculated by the editors from Conseil des prélèvements obligatoires, « Conforter l'égalité des citoyens devant l'imposition des revenus », octobre 2024). Current value: €1,807. Range measured by the source: €1,807 to €2,500; beyond it, linear extrapolation is flagged. Range measured by the two scenarios of the Conseil des prélèvements obligatoires (council on compulsory levies, 2024), up to €2,500. Beyond that, linear extrapolation, flagged as such; complete removal of the cap, costed by the Cour des comptes (€4.5bn to €4.9bn, option), bounds any extrapolation.
  • Abolition of the family quotient in the strict sense (half-shares for children), static: €12.5bn (IGF-IGAS 2026, 2024 tax) to €16.5bn (Cour des comptes, 2021): a gain of €14.5bn per year (12.5 to 16.5).
  • Complete removal of the family quotient cap (abolition of the cap on the benefit per half-share), static: revenue loss costed by the Cour des comptes (2020 income): a cost of €4.9bn per year (4.5 to 4.9).
  • A full share instead of a half-share for the second child (one more half-share for each household with two or more children, ordinary cap), static (editors' calculation): a cost of €2.6bn per year (2.2 to 3) (coefficient calculated by the editors from IGAS et IGF, Revue des dépenses socio-fiscales en faveur de la politique familiale, juillet 2021 (page 41 du PDF)).
  • Abolition of the marital quotient (individual taxation), static: a gain of €11.1bn per year (10 to 11.1).
Current law, horizon, limitations and 9 sources

Current law. Marital quotient (2 shares for a couple) and family quotient (half-share per child, full share from the 3rd). The advantage is capped at €1,807 per half-share in 2026 (2025 income). According to the Cour des comptes (October 2023), taking the family into account in taxation cost €28.3bn in revenue in 2021, of which €16.5bn for the family quotient strictly speaking and €11.1bn for the marital quotient (2021 tax on 2020 income, income tax and CEHR); the IGF and the IGAS (June 2026) put the family quotient at €12.5bn for 2024 (Inès 2024 model). The cap brings in €4.5bn (2021); removing the cap would cost €4.9bn (Cour des comptes).

Horizon. Full year. For 2032, to be updated: the 2014 costing is old, the Cour des comptes costing relates to 2021 tax, and the IGF-IGAS costing to 2024 and 2026.

Limitations. Lowering the ceiling: since 18 October 2026, coefficient derived from the IGF-IGAS (General Inspectorate of Finance and General Inspectorate of Social Affairs) (spending review of June 2026: a 10% cut in the ceiling of €1,807, i.e. about €181, would raise €580m), i.e. €0.32bn per €100 step (editors' calculation, at the centre and as upper bound); lower bound, the costing in the explanatory memorandum of the PLF 2014 (€1.03bn for a ceiling lowered from €2,000 to €1,500, with the share for the first child of single parents lowered by the same amount, 2013 income), i.e. €0.21bn per €100, in 2014 euros (0.20 at the centre before 18 October 2026). It is not linear, because the lower the ceiling, the more households are affected. Abolition of the family quotient in the strict sense: two sources with different methods, as a range (rule of 18 October 2026): €12.5bn for 2024 tax according to the IGF-IGAS (2026, Inès 2024 model; €12.2bn in 2019 according to the IGAS-IGF of 2021), lower bound, and €16.5bn for 2021 according to the Cour des comptes (Court of Audit) (table no. 14), upper bound, estimated from the 2017 amounts of André and Sireyjol (Insee), who measure the family quotient as the difference from joint taxation of couples; midpoint of the range at the centre, €14.5bn (16.5 before 18 October 2026). Marital quotient (quotient conjugal): costing by the Cour des comptes (table no. 14). The Cour des comptes assesses the two quotients one after the other: abolishing only the marital quotient while keeping the family quotient would not give exactly €11.1bn. Complete removal of the ceiling: the Cour des comptes writes that it « représenterait une perte de recettes estimée à 4,9 Md€ pour l'État » [“would represent an estimated revenue loss of €4.9bn for the State”] (2020 income), at the centre and as upper bound, and elsewhere costs the gain from the ceiling in 2021 at €4.5bn, lower bound; the ceiling has since been uprated in line with the tax scale (€1,807 in 2026). Raising the ceiling, extrapolated beyond €2,500 (€0.28bn per €100), remains below this amount up to about €3,550 at the centre. Full share for the second child (option, editors' calculation): the IGAS-IGF (2021) breaks down the €12.2bn of the 2019 family quotient by family size: €3.5bn for 3.1 million one-child households (€1,129 per household, one half-share), €5.4bn for 2.7 million two-child households (€2,000, two half-shares), €3.3bn for about 1 million households with three or more children (€3,300, at least four half-shares). The second half-share is therefore worth on average about €871 (2,000 − 1,129), the third and fourth €650 each ((3,300 − 2,000) / 2): the advantage of a half-share decreases with their number, ceiling included. One more half-share costs at most €871 per two-child household and €650 per larger household (upper bound: 2.7 × 0.871 + 1.0 × 0.650 = €3.0bn), at least €650 and €429 (same decrease of €221; lower bound: €2.2bn); midpoint, €2.6bn. 2019 data, not updated (the family quotient has changed little: €12.5bn in 2024). A microsimulation (LexImpact, OpenFisca) would replace this calculation. Marital quotient: lower bound = static high assumption of the Institut Montaigne (2022), €10bn; with tax optimisation, Montaigne goes down to €7.2bn (not counted, behavioural effect). In practice an abolition would be replaced by a tax credit or a per-child allowance, whose cost would be deducted. Raising the ceiling: coefficient taken from the two scenarios of the CPO (2024), costed on a ceiling of €1,759 (2023 income). €2.1bn for €741 gives €0.28bn per €100 step (central value, over the whole interval); €0.8bn for €241 gives €0.33bn (upper bound): the cost per euro decreases as the ceiling rises. No source read costs a ceiling above €2,500 or a complete removal of the ceiling.

Flat-rate levy on capital (flat tax)costable, revenue
  • Income tax component of the flat tax (PFU). For 1 point increase in the income tax component of the PFU (12.8% to 13.8%), static (derived: €5.3bn / 12.8): a gain of €0.41bn per year (0.35 to 0.41) (coefficient calculated by the editors from Sénat, rapport général n° 108 (2017-2018) sur le PLF 2018, tome II, examen de l'article 11 (prélèvement forfaitaire unique)). Current value: 12.8%. Range measured by the source: 12.8% only; beyond it, linear extrapolation is flagged. Coefficient calculated by the editors from the yield at the current rate (12.8%): any other value is a linear extrapolation, flagged as such.
  • Return to the income tax schedule (end of the PFU): cost of the PFU estimated by the Government assuming unchanged behaviour (€1.9bn in 2019); upper bound, static costing by the Institut Montaigne (2022): a gain of €1.9bn per year (1 to 2.3).
Current law, horizon, limitations and 4 sources

Current law. Flat tax (PFU) of 31.4% since the LFSS 2026: 12.8% income tax and 18.6% social levies (CSG raised to 10.6% for most securities). It was 30% (12.8 + 17.2) from 2018 to 2025. The option for the income tax scale remains open. 2025 yield: €5.3bn of income tax on income and capital gains from securities; €18.5bn of CSG and CRDS on capital income; €14.4bn of other social contributions.

Horizon. Full year, in 2025 euros; to be indexed for 2032.

Limitations. Capital income is highly sensitive to taxation: dividends jumped after 2018. Static costing therefore overstates the gain from an increase; behavioural effects are presented separately. The cost of €1.9bn of the PFU in 2019 (€1.3bn in 2018) is the Government's costing, assuming unchanged behaviour, reported by the Senate (general report on the PLF 2018); the same report cites economists' estimates of « une dizaine de milliards » [“around ten billion”] in steady state, not static and not counted. Upper bound: the Institut Montaigne (2022) costs the return to the income tax schedule at €2.1bn to €2.3bn « à assiette constante » [“on a constant base”] (ex ante cost of the PFU of €1.6bn to €1.8bn, updated with the growth in capital income from 2018 to 2022, in 2022 euros), and at €1.5bn with behavioural effects (not counted). Per-point calculations divide the revenue by the rate, without microsimulation.

Tax expenditures, subsidies and fraud

Tax and social security expenditurescostable, revenue
  • Uniform cut in the cost of tax expenditures. For 1% uniform cut in the total cost of tax expenditures: a gain of €0.88bn per year (0.85 to 1.01) (coefficient calculated by the editors from Fipeco, Les dépenses fiscales). Current value: 0%. Range measured by the source: 0% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total cost of tax expenditures (editors' calculation): arithmetic range from 0 to 100% (abolition of all tax expenditures).
  • Abolition of the tax credit for home employment: a gain of €7.2bn per year (6.4 to 7.2).
  • Abolition of the CIR (research tax credit): a gain of €8.1bn per year (7.7 to 8.1).
  • Return of on-site restaurant dining to the standard VAT rate: a gain of €4.2bn per year.
  • Abolition of the corporate income tax exemption for HLM (social housing) bodies and OPACs (tax expenditure 300102): a gain of €0.34bn per year.
  • Abolition of the corporate income tax exemption for HLM (social housing) bodies, yield fully returned to the landlords (aides à la pierre), zero balance: a gain of €0bn per year.
  • Real estate capital gains excluding main residence: allowance for length of ownership replaced by deduction of inflation alone, at current rates (Cour des comptes, 2025, excluding the effect on sales volume): a gain of €1.9bn per year (1.5 to 3).
  • Income tax exemption for child support received for minor children, the payer keeping their deduction (tax legislation directorate, 2022 income): a cost of €0.12bn per year.
  • Reinstatement of a Pinel-type tax reduction from 2028: cost in 2032 of four to five generations of investments (editors' calculation on the 2019 IGF-CGEDD evaluation): a cost of €1bn per year (0.6 to 1.2) (coefficient calculated by the editors from IGF et CGEDD, Évaluation du dispositif d'aide fiscale à l'investissement locatif Pinel, novembre 2019 (page 22 du PDF)).
  • Abolition of the flat-rate deduction of employer contributions on overtime (social expenditure; all companies since 2026): a gain of €1.02bn per year (0.87 to 1.02) (coefficient calculated by the editors from Assemblée nationale, compte rendu de la 3e séance du 6 novembre 2025 (PLFSS 2026), pages 8943-8944).
  • Abolition of the single tapered general reduction (general reductions in social contributions) on wages from 2 to 3 SMIC, reduction unchanged below that (social expenditure; editors' calculation): a gain of €4.6bn per year (4 to 5.1) (coefficient calculated by the editors from Service-public.fr (entreprendre), Réduction générale dégressive unique (RGDU) des cotisations patronales).
Current law, horizon, limitations and 16 sources

Current law. 474 to about 500 tax expenditures. Total cost in the PLF 2026: €85bn to €88.3bn under the Bercy convention, which counts only the State's share of VAT. 2024 cost: €101.4bn counting all VAT (Fipeco, based on the Cour des comptes report). The most costly: CIR (research tax credit) (€7.7bn to €8.1bn), home employment tax credit (6.4 to 7.2), Dutreil pact (5.0 in 2024), 10% allowance on pensions (4.7 to 4.8), reduced-rate VAT on building works (4.6) and on restaurants (4.2), employee savings (2.9), overtime (2.25 for the income tax portion in 2026, tax expenditure 120146), donations (2.2). Social security expenditures (niches sociales): reductions on low wages €39.1bn in 2025 (merged into the RGDU in 2026); reductions in social contributions about €80bn in total according to the minister (Senate hearing, May 2025). The LPFP 2023-2027 requires −€0.5bn per year of tax expenditures.

Horizon. Full year, 2024-2026 values; not indexed for 2032 (it would follow the base).

Limitations. The costing of a tax expenditure measures the revenue loss assuming unchanged behaviour, not the actual gain from abolishing it (as Bercy itself points out). Totals diverge depending on the VAT convention: €85bn to €88bn (Bercy, PLF 2026) against €101bn (Cour des comptes and Fipeco, 2024). Total of €88.3bn in 2026 and ranking of the most costly tax expenditures: PLF 2026, Voies et moyens, volume II (pages 13 and 25). Social expenditures (reductions, exemptions) are tracked separately, in the PLFSS annex. The 10% across-the-board cut is already given by the linear parameter; « supprimer 23 niches » [“abolish 23 tax expenditures”] is not a defined measure. One measure, one lever: the 10% allowance on pensions is costed under the « abattement-retraites » lever, the Dutreil pact under « droits-succession », the overtime exemption under « heures-supplementaires » (income-tax share: €2.25bn forecast for 2026 in the PLF 2026; the uncompensated social security cost is a different amount, €2.31bn in 2026). A uniform cut overlaps these abolitions, as well as those of this lever's options (research tax credit, home employment, restaurants, social housing bodies (HLM)) and of the other tax expenditures costed elsewhere (employee savings, childcare costs): declared overlaps. Corporate income tax exemption for HLM bodies: €340m forecast for 2026 in the PLF 2026, Bercy's costing assuming unchanged behaviour (reliability rated « bonne », good). The series has been revised downwards: €750m actual in 2023 (PLF 2025), €400m in 2024. An abolition spread over several years is costed at its end point, in full year. PLF 2026 figures: volume II of the Voies et moyens (National Assembly document 1087933). Real estate capital gains (option, 18 October 2026): the Cour des comptes (November 2025) costs replacing the allowance for length of ownership with taking account of monetary erosion alone, at the current rate of 36.2%, at « environ 1,9 Md€, toutefois hors effets volume » [“about €1.9bn, however excluding volume effects”] (2022 transactions), at the centre; upper bound, the difference in its table 13 between the current revenue forgone and that of the « plancher » [“floor”] formula (€2,957m, i.e. « 3 Md€ » [“€3bn”]); lower bound, the €1.9bn scaled to 2023 revenue (€3.6bn instead of 4.7, i.e. −23%), €1.5bn (editors' calculation). The Cour des comptes specifies that these amounts do not measure the budgetary gain, « probablement moindre » [“probably smaller”], because of the fall in sales. Maintenance payments received (option): the tax legislation directorate (direction de la législation fiscale), cited by National Assembly report no. 2121 (November 2025), estimates the exemption at €121m per year (2022 income), for 162,372 taxable households out of 570,000; abolishing the payer's deduction at the same time would raise €590m (not counted in the option). Return of the Pinel scheme (option, editors' calculation): the IGF (General Inspectorate of Finance) and the CGEDD (General Council for the Environment and Sustainable Development) (2019) estimate that at steady state the annual cost equals that of one generation, €2.16bn, for 60,000 dwellings per year, €200,000 on average and a nine-year commitment, i.e. €0.24bn per generation per year; a scheme restored in 2028 has four generations in 2032 (centre, €0.96bn) or five (upper bound, €1.2bn); lower bound, four generations at the reduced rates of 2024 (9% instead of 12% over six years, ratio applied), €0.6bn. The investment volume of a restored scheme is not known (60,000 dwellings: 2016-2017 level). Two social expenditures named by a position (options of 20 October 2026, tax and social security expenditures lever; no other cost schedule option costs them). Flat-rate deduction of employer contributions on overtime (Art. L. 241-18-1 of the code de la sécurité sociale: €1.50 per hour below 20 employees, €0.50 above): €872m forecast for 2026 for companies with fewer than 250 employees, of which €504m below 20 employees and €368m from 20 to 249 (PLFSS 2026, annex 4, table 13, on an accruals basis, before the LFSS measures; €865m in 2024). The LFSS for 2026 (identical amendments no. 879 and 2283, adopted on 6 November 2025) has extended it since 1 January 2026 to companies with 250 employees or more, at a cost of €150m according to the general rapporteur and the minister for public accounts in the chamber. Option « fin-deduction-patronale-heures-sup »: €1.02bn at the centre and as the upper bound (872 + €150m); €0.87bn as the lower bound, the annex amount alone, since the extension has no recorded cost yet. It overlaps neither the income tax exemption nor the reduction in employee contributions on overtime (« heures-supplementaires » lever), which concern other levies. End of the single tapered general reduction (RGDU) between 2 and 3 SMIC (option « fin-rgdu-2-3-smic », editors' calculation): exit point brought down from 3 to 2 SMIC, reduction unchanged below that. No source read publishes the share of the RGDU paid in this band (the “exit at 2 SMIC” scenario of the SMIC Expert Group redesigns the whole curve and is costed relative to the 2025 schedule: it is not the same measure). Calculation: the RGDU coefficient, 0.0200 + Tdelta × [½ × (3 × annual SMIC / annual gross pay − 1)]^1.75, with Tdelta of 0.3781 (fewer than 50 employees) or 0.3821 (50 or more), is worth 5.4% of gross pay at 2 SMIC, 3.7% at 2.25 SMIC, 2.7% at 2.5 SMIC and 2% (floor) up to 3 SMIC; it is applied, band of 0.1 SMIC by band (average rate of each band), to the 2024 private-sector gross wage bill given by the labels of chart I.24 of the SMIC Expert Group (2025 report): amount of the 6-point reduced health rate divided by 0.06 from 2 to 2.4 SMIC (1.2; 1.1; 1.0; €0.9bn), amount of the 1.8-point reduced family rate divided by 0.018 from 2.4 to 3 SMIC (0.3 then €0.2bn per band; the 2.4-2.5 band was only partly eligible for the health rate). Gross wage bill of €142bn between 2 and 3 SMIC (€144bn reading the family rate alone), reduction of €4.56bn (€4.66bn), 4.6 at the centre. Bounds: labels rounded to the nearest tenth of a billion, all taken to within €0.05bn in the same direction, €4.0bn and €5.1bn. 2024 wage bill not indexed, like the neighbouring option “end of reductions above 1.6 SMIC” (the rise in wages from 2024 to 2026 would raise the amount by a few per cent). Static: employment effects, small above 2 SMIC according to the Expert Group, are not counted. Declared overlaps with the abolition of reductions above 1.6 SMIC and with their total abolition (“Reductions in social contributions on low wages” lever), which contain this whole band. « Gel des exonérations à partir de 3 fois le SMIC » [“Freeze of exemptions from 3 times the SMIC”]: since 2026, no general reduction has been paid above 3 SMIC; nothing to cost.

  • Fipeco, Les dépenses fiscales, 2026-01-08: « En 2024, le coût des dépenses fiscales a, pour la première fois, dépassé 100 Md€ pour atteindre 101,4 Md€ (contre 89,4 Md€ dans le PLF avec la nouvelle méthode ...) ... le crédit d'impôt en faveur de la recherche (7,8 Md€), ... l'exonération partielle de droits de mutation en cas de transmission d'entreprises (5,0 Md€), l'abattement de 10 % sur le montant des pensions et retraites (4,8 Md€) »
  • PLF 2026, Évaluation des voies et moyens, tome II (dépenses fiscales), pages 13 et 25 du PDF, 2025-10: « Le montant des dépenses fiscales s’est élevé à 89,4 Md€ en 2024 (montant définitif, +6,1 Md€ par rapport aux prévisions) et s’élèverait à 91,8 Md€ en 2025 (+6,7 Md€ par rapport aux prévisions) et 88,3 Md€ en 2026. […] 1 200302 Crédit d’impôt en faveur de la recherche 8 041 2 110246 Crédit d’impôt au titre de l’emploi d’un salarié à domicile 7 208 3 120401 Abattement de 10 % sur le montant des pensions (y compris les pensions alimentaires) et des retraites 4 665 »
  • Fipeco, Les allègements de cotisations sociales patronales sur les bas salaires, 2026-09-24: « Leur coût total était de 39,1 Md€ en 2025 ... à partir de 2026, ces allégements ... sont fusionnés en une « réduction générale dégressive unique » »
  • PLF 2026, Évaluation des voies et moyens, tome II, Dépenses fiscales, 2025-10: « 300102 – Exonération des organismes d'HLM et des offices publics d'aménagement et de construction (OPAC) ; Finalité : Aider le secteur immobilier social ; Réalisation 2024 : 400 ; Prévision 2025 : 340 ; Prévision 2026 : 340 (en millions d'euros) ; Bénéficiaires : 659 Entreprises ; Fiabilité du réalisé : Bonne ; Référence : code général des impôts 207-1-4°, 221 bis »
  • PLF 2025, Évaluation des voies et moyens, tome II, 2024-10: « 300102 – Exonération des organismes d'HLM et des offices publics d'aménagement et de construction (OPAC) ; Réalisation 2023 : 750 ; Prévision 2024 : 400 ; Prévision 2025 : 400 »
  • Cour des comptes, « L'imposition des plus-values immobilières des particuliers », novembre 2025 (pages 9, 70 et 81), 2025-11: « S'agissant des résidences secondaires, le mécanisme d'abattement progressif actuel correspond à un manque à gagner pour les finances publiques de 3 Md€ par rapport à une formule « plancher » qui consisterait à prendre en compte la seule érosion monétaire […] Tableau n° 13 : différence de rendement en 2022 entre le dispositif actuel et un dispositif intégrant l'érosion monétaire (en M€) […] 4 854 […] 1 897 […] Différence 2 957 […] Si cette substitution se fait sur la base du taux actuel d'imposition forfaitaire de 36,2 %, le surcroît d'imposition serait de 42 % rapporté aux transactions de 2022, année où le rendement de l'impôt avait été de 4,5 Md€. Cette mesure pourrait donc rapporter environ 1,9 Md€, toutefois hors effets volume sur le nombre de transactions. »
  • Cour des comptes, même rapport, synthèse (page 8 du PDF), 2025-11: « Alors que ce produit atteignait 4,7 Md€ en 2022, soit 72 % de plus que le produit observé en 2017, le retournement du marché a entraîné une diminution de 23 % des recettes en 2023 (3,6 Md€) »
  • Assemblée nationale, rapport n° 2121 (commission des affaires sociales), 19 novembre 2025, 2025-11-19: « s’agissant de l’exonération d’impôt non plafonnée sur la pension alimentaire perçue par les parents créanciers pour leurs enfants mineurs, sur la base des revenus 2022, la direction de la législation fiscale estime le coût de la mesure à 121 millions d’euros par an, réduisant l’impôt sur le revenu des 162 372 foyers fiscaux imposables parmi les 570 000 concernés […] s’agissant de la suppression de l’abattement fiscal sur la pension alimentaire versée par les parents débiteurs pour leurs enfants mineurs, la direction de la législation fiscale estime qu’elle permettrait un gain de 590 millions d’euros annuels »
  • IGF et CGEDD, Évaluation du dispositif d'aide fiscale à l'investissement locatif Pinel, novembre 2019 (page 22 du PDF), 2019-11: « La mission a estimé par ailleurs, qu'en régime de croisière, le coût pour l'État serait égal au coût d'une génération, soit 2,16 Md€ par an, en partant d'une hypothèse conservatrice sur le nombre et le coût moyen des logements. […] 60 000 logements par an, soit un niveau intermédiaire entre 2016 et 2017, un montant moyen de 200 000 € et un engagement de location pendant neuf ans. »
  • PLFSS 2026, annexe 4, tableau 13 (page 45 du PDF) : évolution tendancielle 2024-2026 du coût des allègements généraux et des exonérations ciblées compensées, champ ROBSS, 2025-10: « En M€, en droits constatés ; 2024 ; 2025 (P) ; Évolution 2024-2025 ; 2026 (P) ; Évolution 2025-2026 […] Déductions sur les heures supplémentaires (entreprises de moins 250 salariés) 865 868 0,4% 872 0,5% ; Dont entreprises d'au moins 20 et de moins de 250 salariés 360 364 1,1% 368 1,2% ; Dont entreprises de moins de 20 salariés 505 505 -0,1% 504 -0,1% […] SOURCE : COMMISSION DES COMPTES DE LA SECURITE SOCIALE (CCSS) D'OCTOBRE 2025 ET DONNEES TRANSMISES PAR LES REGIMES, AVANT MESURES LFSS »
  • Assemblée nationale, compte rendu de la 3e séance du 6 novembre 2025 (PLFSS 2026), pages 8943-8944, 2025-11-06: « (Les amendements identiques nos 879 et 2283, modifiés par la suppression du gage, sont adoptés.) […] Mme Amélie de Montchalin, ministre. […] Comme le rapporteur général, j'ai bien indiqué que l'amendement no 879 de la Droite républicaine avait un coût de 150 millions d'euros. »
  • Assemblée nationale, compte rendu de la 2e séance du 6 novembre 2025 (PLFSS 2026), pages 8888-8889, 2025-11-06: « M. Thibault Bazin, rapporteur général. […] Votre autre amendement prévoit une déduction forfaitaire patronale dans les entreprises de plus de 250 salariés. Cette mesure permettrait d'inciter aux heures supplémentaires les structures de cette taille et elle aurait un coût plus acceptable pour notre trajectoire – 150 millions d'euros. »
  • LégiSocial, Réduction loi TEPA 2026 (heures supplémentaires) : extension de la déduction patronale par la LFSS pour 2026, 2026: « Depuis le 1er janvier 2026, la LFSS pour 2026 a étendu la déduction patronable aux employeurs d'au moins 250 salariés. […] Déduction forfaitaire patronale à hauteur de 0,50€/ heure supplémentaire. »
  • Service-public.fr (entreprendre), Réduction générale dégressive unique (RGDU) des cotisations patronales, 2026: « Coefficient = Tmin + (Tdelta × [(1/2) × (3 × SMIC calculé pour un an, valeur au 1er janvier 2026 / rémunération annuelle brute - 1)]P) ; Tmin : 0,0200 ; Tdelta : 0,3781 pour moins de 50 salariés, 0,3821 à partir de 50 salariés ; P : 1,75 […] L'allègement de cotisations ne s'applique plus à partir d'une rémunération égale à 3 Smic. »
  • Groupe d'experts SMIC, Rapport 2025, graphique I.24 (p. 70) : coût des allègements généraux par tranche de salaire en part de SMIC, 2024, 2025-11-25: « Étiquettes lues (Md€), tranches [2,0-2,1[ à [2,4-2,5[ : « Montant taux réduit AM » 1,2 ; 1,1 ; 1,0 ; 0,9 ; 0,5 ; « Montant taux réduit AF » 0,4 ; 0,3 ; 0,3 ; 0,3 ; 0,3 ; tranches [2,5-2,6[ à [2,9-3,0[ : « Montant taux réduit AF » 0,2 ; 0,2 ; 0,2 ; 0,2 ; 0,2. Champ : secteur privé au sens de l'Urssaf-Caisse nationale. Source : DSS/EPF/6C, sur données Urssaf-Caisse nationale »
  • Groupe d'experts SMIC, Rapport 2025, chapitre IV (p. 137) : profil de la RGDU, 2025-11-25: « la nouvelle formule issue du décret du 4 septembre 2025 se traduit par un lissage progressif de la courbe et une réduction des taux d'allègement pour les salaires supérieurs à environ 1,8 Smic, avec une extinction complète du dispositif autour de 3 Smic »
Public aid to businessescostable, spending
  • Cut in budgetary spending on business support (excluding reductions in social contributions and tax expenditures). For 1% cut in the €30bn of budgetary spending on business support (minister's breakdown, Senate report): a gain of €0.3bn per year (coefficient calculated by the editors from Sénat, rapport n° 808 (2024-2025), synthèse, Aides publiques aux entreprises). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 2 sources

Current law. There is no legal definition and no regularly updated official costing. The Senate commission of inquiry (report No. 808, July 2025) estimates aid at « au moins 211 milliards d'euros en 2023 » [“at least 211 billion euros in 2023”] in the broad sense, and at €108bn excluding Bpifrance, reclassified tax expenditures and VAT. The minister at a hearing: €150bn (40 in tax expenditures, 30 in budget spending, 80 in reductions in social contributions). In national accounts terms, non-financial corporations received €73bn in aid in 2024 (Fipeco).

Horizon. Full year; 2023-2025 values.

Limitations. The scope is decisive: it makes the amount vary from €73bn to €211bn. The cost schedule uses only one, without overlap: the €30bn of budgetary spending in the minister's breakdown. Reductions in social contributions (€80bn) are costed under the “Reductions in social contributions on low wages” lever, tax expenditures (€40bn) under the “Tax and social security expenditures” lever. Changing the reductions has an effect on employment; static costing does not measure it. Making aid conditional (counterparts) cannot be costed as such. The 10% cut is given by the linear parameter; the abolition of the CIR (research tax credit) is under the « niches-fiscales » lever. Reduction announced as an amount without a list of the aid targeted: a pure amount with no named measure, not costed, cited next to the total (decision 1 of 18 October 2026, common rule), like the cut in production taxes announced as an amount (“Production taxes” lever).

Tax and social security fraudcostable, revenue
  • Increase in revenue recovered through tax audits. For +10% in tax audit revenue (base €11.4bn recovered; €20.1bn assessed for collection): a gain of €1.14bn per year (1 to 2) (coefficient calculated by the editors from Cour des comptes, La lutte contre la fraude fiscale). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • +10% recovery of social fraud (base €0.6bn): a gain of €0.06bn per year (coefficient calculated by the editors from Cour des comptes, La lutte contre la fraude fiscale).
  • Recover 10% of estimated social fraud (€13bn): a gain of €1.3bn per year (0.6 to 1.3) (coefficient calculated by the editors from Cour des comptes, La lutte contre la fraude fiscale).
Current law, horizon, limitations and 2 sources

Current law. Tax audits in 2024: €20.1bn of duties and penalties assessed for collection, €11.4bn actually collected (Cour des comptes, December 2025). No official costing of tax fraud: for the Cour, « nul ne peut aujourd'hui se hasarder à affirmer que ce montant est proche de 30 ou de 130 Md€ » [“no one today can venture to assert that this amount is close to €30bn or €130bn”]. VAT gap: €6bn to €10bn (DGFiP). Social security fraud estimated at €13bn, of which €2.1bn detected and €0.6bn recovered (HCFiPS, 2024). Anti-fraud bill under discussion in 2026.

Horizon. Phased in over several years; full year 2032 to be justified by resources (staff, tools).

Limitations. The amounts of « fraude totale » [“total fraud”] often cited (€80bn to €100bn, Solidaires) are judged by the Cour des comptes to have no method. A promise of yield from the fight against fraud is costed on the revenue recovered, not on the estimated fraud, and related to the resources committed. Coefficient calculated by the editors (10% of the €11.4bn recovered; €20.1bn issued for collection as the upper bound). The costing is fragile: low confidence index.

  • Cour des comptes, La lutte contre la fraude fiscale, 2025-12-16: « Celles-ci sont passées de 12,2 Md€ en 2015 à 11,4 Md€ en 2024 ... ces résultats retrouvent leur niveau de départ (20,1 Md€ en 2024 contre 21,2 Md€ en 2015) ... Nul ne peut aujourd'hui se hasarder à affirmer que ce montant est proche de 30 ou de 130 Md€ ... La fourchette finalement dégagée par la DGFiP (entre 6 et 10 Md€ ...) »
  • HCFiPS, Fraude sociale 2024, infographie, 2024-09-25: « MONTANTS ÉVALUÉS ... 13 MD€ ; FRAUDES CONSTATÉES ET STOPPÉES 2,1 MD€ DONT 0,5 MD€ ÉVITÉES ; MONTANTS RECOUVRÉS 0,6 MD€ »

Wealth and inheritance

Wealth tax (IFI, ISF)costable, revenue
  • Reinstate the wealth tax (ISF) (2017 base): revenue loss from its abolition, mechanical effect (Conseil d'analyse économique): a gain of €3.75bn per year (3 to 4.5).
  • Tax on the financial assets of holding companies (PLF 2026): a gain of €1bn per year.
  • Minimum tax of 2% on fortunes above €100m (« taxe Zucman », private member's bill passed by the National Assembly in February 2025), static: high estimate, that of its proponents (Gabriel Zucman, National Assembly report no. 930 and Senate report no. 689); the administration has not simulated it; estimates that take account of exile, tax optimisation or the deduction of corporate income tax are much lower (€2bn to €5bn): a gain of €20bn per year (15 to 25).
Current law, horizon, limitations and 7 sources

Current law. IFI (real estate wealth tax, since 2018): net real estate assets above €1.3m; scale from 0.5% to 1.5% above €0.8m; capped at 75% of income. 194,000 tax assessments in 2025, for €2.3bn in budget accounting (€2.7bn in national accounts); €3.1bn forecast for 2026. The ISF raised €5.1bn in 2017 (358,000 households). The PLF 2026 creates a tax on the financial assets of holding companies (+€1.0bn forecast). No minimum tax on very large fortunes (« taxe Zucman »): the private member's bill adopted by the National Assembly in February 2025 was rejected by the Senate in June 2025.

Horizon. Full year; ISF counterfactual in 2022 euros, not updated for 2032; minimum tax estimated on 2024 fortunes (Challenges ranking), not updated.

Limitations. Restoring the wealth tax (ISF): central value, €3.75bn, midpoint of the range of the Conseil d'analyse économique (Council of Economic Analysis, 2025), which puts the revenue loss from abolishing the ISF at between €3bn and €4.5bn, mechanical effect only (lower and upper bounds). Benchmarks: €3.2bn, ex ante revenue loss from converting the ISF into the IFI according to the impact assessment of the PLF 2018 (Senate general report); €4.5bn net from the IFI according to a press summary of the France Stratégie evaluation committee (2022 counterfactual). The 2026 yield of the tax on holding companies depends on the text adopted; the ISF and this tax partly bear on the same assets (declared overlap). Minimum tax of 2% on wealth above €100m (« taxe Zucman »): costed and added at the upper end of the estimates, €15bn to €25bn (about 20, « avec une marge d'erreur d'environ 5 milliards » [“with a margin of error of about 5 billion”] according to Public Sénat), a static estimate by its promoters alone (the economists heard, including Gabriel Zucman), taken up in the reports of the National Assembly (no. 930) and the Senate (no. 689); it is the only static estimate read of the scheme adopted by the National Assembly in February 2025. The administration has not simulated it: the DGFiP stated that it could not do so, for lack of a valuation of financial and business assets. The estimate starts from the Challenges ranking (€1,228bn for the 500 largest fortunes in 2024) and a current tax rate of about 0.3% of wealth. Estimates that take into account exile, optimisation or the deduction of corporate income tax are much lower, €2bn to €5bn: if corporate income tax were deducted, the yield would fall to €2bn or €3bn according to Antoine Levy (cited by the Senate); the figures of about €5bn in circulation take into account exile and optimisation, or this deduction; according to the Senate report, the departure of the three largest fortunes would halve the yield; the Institut de l'entreprise (2026) speaks of a yield « divisé par quatre » [“divided by four”]. CAE Focus no. 118 does not cost this tax (it simulates tax exile for tax increases of €2bn and €4bn). The amount counted is therefore the top of the range, to be read as such. As a differential tax « tous impôts confondus » [“all taxes combined”], it would yield slightly less if the ISF were restored, since the ISF paid on the same assets would be deducted; no source read measures this common share, and the two amounts are added together. A floor limited to billionaires is not costed.

Inheritance and gift taxcostable, revenue
  • Change in the yield of inheritance and gift tax. For 1% change in the yield of inheritance and gift tax (1% of €21.2bn in 2025): a gain of €0.21bn per year (coefficient calculated by the editors from Projet de loi de finances rectificative pour 2012 (Assemblée nationale, n° 71), évaluations préalables, article 4 (pages 172 et 173 du PDF)). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • Direct-line allowance, per child and per parent (inheritances and gifts). For +€10,000 on the direct-line allowance (derived from the impact assessment of the PLFR 2012: €1,325m for an allowance lowered from €159,325 to €100,000, at cruising speed in 2014): a cost of €0.22bn per year (coefficient calculated by the editors from Projet de loi de finances rectificative pour 2012 (Assemblée nationale, n° 71), évaluations préalables, article 4 (pages 172 et 173 du PDF)). Current value: €100,000. Range measured by the source: €100,000 to €159,325; beyond it, linear extrapolation is flagged. Average cost measured between €100,000 and €159,325 (2012 simulation). Beyond that, linear extrapolation, flagged as such.
  • Direct-line allowance on inheritances only, per child and per parent. For +€10,000 on the direct-line allowance for inheritances (derived from the impact assessment of the PLFR 2012: €895m of inheritance duties for an allowance lowered from €159,325 to €100,000, at cruising speed in 2014): a cost of €0.15bn per year (coefficient calculated by the editors from Projet de loi de finances rectificative pour 2012 (Assemblée nationale, n° 71), évaluations préalables, article 4 (pages 172 et 173 du PDF)). Current value: €100,000. Range measured by the source: €100,000 to €159,325; beyond it, linear extrapolation is flagged. Average cost measured between €100,000 and €159,325 (2012 simulation). Beyond that, linear extrapolation, flagged as such.
  • Direct-line allowance on gifts only, per child and per parent. For +€10,000 on the direct-line allowance for gifts (derived from the impact assessment of the PLFR 2012: €430m of gift duties for an allowance lowered from €159,325 to €100,000, at cruising speed in 2014): a cost of €0.07bn per year (coefficient calculated by the editors from Projet de loi de finances rectificative pour 2012 (Assemblée nationale, n° 71), évaluations préalables, article 4 (pages 172 et 173 du PDF)). Current value: €100,000. Range measured by the source: €100,000 to €159,325; beyond it, linear extrapolation is flagged. Average cost measured between €100,000 and €159,325 (2012 simulation). Beyond that, linear extrapolation, flagged as such.
  • Exemption rate on shares transferred under a Dutreil pact. For 1 more point of exemption (derived from the CPO (council on compulsory levies) 2025: bringing the exemption down from 75% to 50% would reduce the tax expenditure by €1.3bn; upper bound, tax expenditure of €5.0bn in 2024 divided by the 75 points): a cost of €0.052bn per year (0.052 to 0.067) (coefficient calculated by the editors from Conseil des prélèvements obligatoires, « Corriger les principales distorsions de l'imposition du patrimoine », décembre 2025 (page 209 imprimée)). Current value: 75%. Range measured by the source: 50% to 75%; beyond it, linear extrapolation is flagged. Range measured by the CPO (2025): from 50 to 75%. Beyond 75%, linear extrapolation, flagged as such.
  • Abolition of the pacte Dutreil (2024 tax expenditure): a gain of €5bn per year (3 to 5).
  • Abolition of inheritance tax (excluding gifts): a cost of €17bn per year (16.1 to 18.1).
  • Abolition of direct-line inheritance tax (children), excluding gifts: 53% of inheritance tax (DG Trésor model, 2018) applied to current revenue (editors' calculation): a cost of €9bn per year (8.5 to 9.6) (coefficient calculated by the editors from Cour des comptes, « Les droits de succession », septembre 2024, annexe n° 7 (page 92)).
  • 100% levy on the part of transfers received above €12m per heir, static (editors' calculation on the CAE distribution of inheritances; upper bound: Institut Montaigne, 2022, threshold assessed on net taxable wealth): a gain of €3.8bn per year (3.5 to 9) (coefficient calculated by the editors from CAE, note n° 69, « Repenser l'héritage », décembre 2021 (page 3)).
  • Allowance on direct-line inheritances raised from €100,000 to €150,000 per heir (5 steps of the « successions » [“inheritances”] parameter, within the range measured): a cost of €0.75bn per year (coefficient calculated by the editors from Projet de loi de finances rectificative pour 2012 (Assemblée nationale, n° 71), évaluations préalables, article 4 (pages 172 et 173 du PDF)).
  • Dutreil exemption raised from 75 to 90% (15 steps of the « pacte Dutreil » [“Dutreil pact”] parameter, extrapolated beyond the range measured by the CPO): a cost of €0.78bn per year (0.78 to 1) (coefficient calculated by the editors from Conseil des prélèvements obligatoires, « Corriger les principales distorsions de l'imposition du patrimoine », décembre 2025 (page 209 imprimée)).
  • Allowance on direct-line gifts raised from €100,000 to €200,000 (10 steps of the « donations » [“gifts”] parameter, extrapolated beyond the range measured): a cost of €0.7bn per year (coefficient calculated by the editors from Projet de loi de finances rectificative pour 2012 (Assemblée nationale, n° 71), évaluations préalables, article 4 (pages 172 et 173 du PDF)).
  • Contribution of 10% on the part of transfers received above €4m per heir, and 10% more above €13m, static (editors' calculation on the CAE distribution of inheritances): a gain of €1.86bn per year (1.8 to 1.92) (coefficient calculated by the editors from CAE, note n° 69, « Repenser l'héritage », décembre 2021 (page 3)).
Current law, horizon, limitations and 10 sources

Current law. In the direct line: allowance of €100,000 per child and per parent, gifts recalled for tax purposes over 15 years, scale from 5% to 45% (20% from €15,900 to €552,000; 45% above €1,806,000). 60% with no family relationship. Life insurance: €152,500 per beneficiary, then 20% and 31%. Dutreil pact: 75% exemption on company shares. DMTG (inheritance and gift duties): €21.2bn in 2025, of which €16.1bn from inheritances; €23.0bn forecast for 2026, of which €18.1bn from inheritances (including exceptional transactions), and €22.0bn in 2027, of which €17.0bn from inheritances (Voies et moyens, PLF 2027).

Horizon. Full year; the yield grows with property and asset prices; not indexed for 2032.

Limitations. Allowance in the direct line: the impact assessment of article 4 of the PLFR 2012 (amending finance bill) costed lowering it from €159,325 to €100,000 at €895m of inheritance tax and €430m of gift duties per year from 2014 (microsimulation on the 2006 inheritance survey, static); hence €0.22bn per €10,000 step (€1,325m / €59,325), in 2014 euros, average cost over this interval. The Institut Montaigne (2022), which cites this study, writes €825m and retains inheritances only (€0.15bn per step, about €0.2bn updated at 4% per year): the source says €895m. The coefficient is not updated: no source read brings it to 2026 or 2032. Beyond €159,325 (€200,000, for example), no official costing read; nor of a « héritage maximum » (maximum inheritance): a microsimulation based on DGFiP inheritance data would be needed. Abolishing the Dutreil pact would run up against avoidance strategies: the figure is the cost of the tax expenditure, not a guaranteed gain. Abolition of inheritance tax: cost equal to its revenue, excluding gifts, according to the Voies et moyens of the PLF 2027: €16.1bn in 2025 (lower bound), €17.0bn forecast for 2027 (central value) and €18.1bn forecast for 2026, including exceptional transactions (upper bound). The “1% of revenue” coefficient is calculated by the editors (1% of €21.2bn). The abolition overlaps a change in revenue and the Dutreil pact (declared overlaps). This static costing ignores the shift towards inheritances of transfers currently made by gift. Separate allowances (parameters, 18 October 2026): the same 2012 assessment separates inheritances (€895m, i.e. €0.15bn per €10,000 step) and gifts (€430m, €0.07bn), in 2014 euros. Dutreil pact (parameter, 18 October 2026): the CPO (December 2025) costs at €1.3bn the reduction in the tax expenditure if the exemption went from 75% to 50%, i.e. €0.052bn per point (centre); upper bound, the tax expenditure of €5.0bn (2024) divided by the 75 points (0.067); an increase beyond 75% is extrapolated. Abolition in the direct line (option, editors' calculation): according to the DG Trésor microsimulation model published by the Cour des comptes (2024, annex 7), children paid €7.1bn of the €13.4bn of inheritance tax in 2018, i.e. 53%; applied to the €16.1bn / €17.0bn / €18.1bn of revenue (Voies et moyens, PLF 2027), €8.5bn / €9.0bn / €9.6bn. Large inheritances (options, editors' calculation): the CAE (Council of Economic Analysis) (note no. 69, 2021) estimates that within a generation, the top 1% of heirs receive on average more than €4.2m net of duties and the top 0.1% about €13m; its « capital pour tous » [“capital for all”] scenarios (€9,600 per person for €12.3bn, €7,200 for €9.3bn) imply about 1.29 million heirs per generation, taken as an annual flow (calculation). A Pareto distribution fitted to these two averages (coefficient of 2.04 to 2.26, top 0.1% threshold of €6.6m to €7.3m) gives the amount received each year above a threshold: €3.9bn to €4.4bn above €12m, €13.9bn to €15.7bn above €4m, €3.6bn to €4.1bn above €13m. 100% levy above €12m: this amount minus the duties already paid (about 10% at the top of the distribution, CAE), i.e. €3.5bn to €4.0bn (3.8 at the centre); upper bound, the median estimate of the Institut Montaigne (2022), « de l'ordre de 9 milliards d'euros » [“of the order of 9 billion euros”], for a threshold assessed on net taxable wealth. Contribution of 10% above €4m and a further 10% above €13m: 10% of each of the two amounts, €1.80bn to €1.92bn. Inheritances accumulated over a lifetime (CAE definition), gross economic wealth, exemptions included: on taxable assets alone, the amount would be lower; data from 2006 to 2018, not updated; no behavioural effect (early gifts, expatriation), which is very strong at these levels: static costings, to be read as upper limits.

  • Projet de loi de finances rectificative pour 2012 (Assemblée nationale, n° 71), évaluations préalables, article 4 (pages 172 et 173 du PDF), 2012-07-04: « La simulation a consisté à abaisser à 100 000 € l’abattement en ligne directe de 159 325 € prévu au I de l’article 779 du CGI en législation 2012. Compte tenu des délais en matière de succession, le rendement en régime de croisière est obtenu en 2014. Il est estimé à : - 10 M€ en 2012, 700 M€ en 2013 et 895 M€ à compter de 2014 au titre des droits de succession ; - 130 M€ en 2012, 440 M€ en 2013 et 430 M€ à compter de 2014 au titre des droits de donation. »
  • Fipeco, Les impôts sur le patrimoine des ménages, 2026: « Les droits de mutation à titre gratuit (DMTG) ont rapporté 21,1 Md€ en 2025, dont 16,1 Md€ pour les droits de successions ... en ligne directe : 5 % jusqu'à 8 100 € ; ... 20 % de 15 900 à 552 000 € ; 30 % de 552 000 à 902 000 € ; 40 % de 902 000 à 1 806 000 € et 45 % au-delà. Sans lien de parenté, le taux est de 60 % »
  • PLF 2026, Voies et moyens tome I, 2025-10: « La prévision s'établit à 21,4 Md€ pour 2026 ... Mutations à titre gratuit par décès (successions - ligne 1706) : 17,0 Md€ en 2026 »
  • Fipeco, Les dépenses fiscales, 2026-01-08: « l'exonération partielle de droits de mutation en cas de transmission d'entreprises (5,0 Md€) »
  • PLF 2027, Évaluation des voies et moyens, tome I (page 42), 2026-10: « Les recettes se sont élevées à 21,2 Md€ en 2025 et se répartissent ainsi : […] Mutations à titre gratuit par décès (successions ligne 1706) : 16,1 Md€ en 2025 […] La prévision s’établirait à 23,0 Md€ pour 2026, dont 4,8 Md€ pour les mutations à titre gratuit (DMTG) entre vifs (donations - ligne 1705), en hausse de +0,4 Md€, et 18,1 Md€ pour les mutations à titre gratuit par décès (successions - ligne 1706) […] Mutations à titre gratuit par décès (successions - ligne 1706) : 17,0 Md€ en 2027 »
  • Cour des comptes, « Les droits de succession », septembre 2024, annexe n° 7 (page 92), 2024-09: « Annexe n° 7 : caractéristiques des droits de succession en fonction du lien de parenté, selon le modèle de micro-simulation de la direction du Trésor (2018) […] Droits de succession (Md€) 0 7,1 2,3 1,1 1,2 1,4 0,4 13,4 […] Note de lecture : en 2018, l'héritage transmis aux enfants s'élevait à 88,4 Md€ pour un total de patrimoine transmis par succession de 135,8 Md€. Le montant des droits de succession acquitté par les héritiers en ligne directe, soit les enfants de la personne décédée, est de 7,1 Md€, soit un taux moyen de taxation de 8,0 %. »
  • CAE, note n° 69, « Repenser l'héritage », décembre 2021 (page 3), 2021-12: « Au sein d'une cohorte, 50 % des individus auront hérité de moins de 70 000 euros de patrimoine tout au long de leur vie […] le top 1 % des héritiers d'une génération recevra en moyenne plus de 4,2 millions d'euros nets de droits et le top 0,1 % environ 13 millions d'euros. […] le top 0,1 % de chaque cohorte, qui aura reçu au cours de la vie environ 13 millions d'euros de transmissions brutes, ne paie qu'à peine 10 % de droits de succession sur l'ensemble de ce patrimoine hérité »
  • CAE, Focus n° 077, « Repenser l'héritage : analyses supplémentaires », décembre 2021, tableau 12, 2021-12: « Tableau 12. Propositions de transferts de patrimoines pour chaque scénario de réforme de la taxation sur l'héritage […] Demogrant 7 200 euros à P0-100 […] 9 600 euros à P0-100 […] Recettes fiscales supplémentaires totales (en milliards d'euros) 9,3 19,0 12,3 »
  • Institut Montaigne, « Présidentielle 2022 – Chiffrage : Quel(s) impôt(s) pour les successions et les donations ? », reproduit par Le Monde du Droit, 2 février 2022, 2022-02-02: « La mesure du candidat de la France Insoumise consisterait, à minima, à relever ou créer une nouvelle tranche marginale d'imposition des successions dont le taux serait de 100 % au-delà de 12 millions d'euros de patrimoine net taxable. […] Si elle était mise en œuvre, la mesure générerait, selon l'estimation médiane de l'Institut Montaigne, un surcroît de recettes de l'ordre de 9 milliards d'euros. »
  • Conseil des prélèvements obligatoires, « Corriger les principales distorsions de l'imposition du patrimoine », décembre 2025 (page 209 imprimée), 2025-12: « L'abaissement du taux d'exonération partielle prévu par le pacte Dutreil de 75 % à 50 % […] réduirait la dépense fiscale d'un montant estimé à 1,3 Md€. »
Property taxcostable, revenue
  • Change in the yield of property tax on built land. For 1% change in the yield of the TFPB (property tax on built land) (households only 0.27 / total €0.43bn, derived): a gain of €0.43bn per year (0.27 to 0.43) (coefficient calculated by the editors from Fipeco, Les impôts locaux). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • Abolition of the household share, compensated by the State to municipalities (derived, ceiling): a cost of €27bn per year (coefficient calculated by the editors from Fipeco, Les impôts locaux).
Current law, horizon, limitations and 2 sources

Current law. Property tax on built property: rate voted by municipalities and intermunicipal bodies (and no longer by the départements since 2021), applied to the cadastral income (half the rental value, 1970s bases uprated). Yield: €43.1bn in 2025, plus €1.2bn on unbuilt land, of which €27.0bn paid by households.

Horizon. Full year; local revenue. A cut decided by the State only costs the State if it compensates.

Limitations. No source read officially breaks down revenue between main residences and households' other dwellings; an exemption for main residences requires data from the DGFiP or the OFGL. Coefficients calculated by the editors (1% of revenue). The revision of the rental values of dwellings has been postponed; its effect is redistributive and revenue-neutral if rates adjust.

Consumption

VATcostable, revenue
  • Standard VAT rate. For 1 point on the standard rate (20% to 21%): a gain of €7.5bn per year (7.5 to 8.9). Current value: 20%. Range measured by the source: 19% to 21%; beyond it, linear extrapolation is flagged. The DG Trésor costs the yield of one rate point; beyond a one-point change, linear extrapolation, flagged as such.
  • Intermediate VAT rate (10%). For 1 point on the 10% rate (2025 net yield as central value, gross as upper bound): a gain of €1.6bn per year (1.6 to 1.9). Current value: 10%. Range measured by the source: 9% to 11%; beyond it, linear extrapolation is flagged. The DG Trésor costs the yield of one point of this rate (tables 1 and 2); beyond a one-point change, linear extrapolation, flagged as such, which ignores behavioural effects and shifts in consumption.
  • Reduced VAT rate (5.5%). For 1 point on the 5.5% rate (2025 net yield as central value, gross as upper bound): a gain of €2bn per year (2 to 2.4). Current value: 5.5%. Range measured by the source: 4.5% to 6.5%; beyond it, linear extrapolation is flagged. The DG Trésor costs the yield of one point of this rate (tables 1 and 2); beyond a one-point change, linear extrapolation, flagged as such, which ignores behavioural effects and shifts in consumption.
  • Special VAT rate (2.1%). For 1 point on the 2.1% rate (2025 yield, net and gross equal at the rounding of the publication): a gain of €0.4bn per year. Current value: 2.1%. Range measured by the source: 1.1% to 3.1%; beyond it, linear extrapolation is flagged. The DG Trésor costs the yield of one point of this rate (tables 1 and 2); beyond a one-point change, linear extrapolation, flagged as such, which ignores behavioural effects and shifts in consumption.
  • +1 point on the 10% rate: a gain of €1.6bn per year (1.6 to 1.9).
  • +1 point on the 5.5% rate: a gain of €2bn per year (2 to 2.4).
  • +1 point on the 2.1% rate: a gain of €0.4bn per year.
  • +1 point on all rates (net of general government; gross as upper bound): a gain of €11.4bn per year (11.4 to 13.7).
  • VAT at 5.5% instead of 20% on motor fuels, electricity, gas and heating oil (static, 2022 data): a cost of €9.52bn per year (9.52 to 11.3) (coefficient calculated by the editors from DG Trésor, Trésor-Éco n° 371, « Analyse de la composition des recettes de TVA » (M. Gesta), septembre 2025, tableaux 1 et 2 (pages 4 et 5 du PDF)).
  • VAT at 5.5% instead of 20% on natural gas and electricity only (static, 2022 data): a cost of €3.15bn per year (3.15 to 3.74) (coefficient calculated by the editors from DG Trésor, Trésor-Éco n° 371, « Analyse de la composition des recettes de TVA » (M. Gesta), septembre 2025, tableaux 1 et 2 (pages 4 et 5 du PDF)).
  • VAT at 5.5% instead of 20% on motor fuels and electricity only (static, 2022 data): a cost of €8.31bn per year (8.31 to 9.86) (coefficient calculated by the editors from DG Trésor, Trésor-Éco n° 371, « Analyse de la composition des recettes de TVA » (M. Gesta), septembre 2025, tableaux 1 et 2 (pages 4 et 5 du PDF)).
  • VAT at 5.5% instead of 20% on motor fuels, electricity and gas, excluding heating oil (static, 2022 data): a cost of €9.09bn per year (9.09 to 10.79) (coefficient calculated by the editors from DG Trésor, Trésor-Éco n° 371, « Analyse de la composition des recettes de TVA » (M. Gesta), septembre 2025, tableaux 1 et 2 (pages 4 et 5 du PDF)).
  • Home improvement and maintenance work (excluding energy renovation) at the standard rate of 20% instead of 10%: cost of tax expenditure 730213 (Voies et moyens, PLF 2026), static: a gain of €2.47bn per year (2.2 to 2.47).
  • End of the VAT charged on the excise duty on motor fuels (« TVA sur la taxe » [“VAT on the tax”]), share paid by households (businesses recover their VAT), static (editors' calculation): a cost of €4bn per year (3.8 to 4) (coefficient calculated by the editors from PLF 2027, Évaluation des voies et moyens, tome I, accises sur les carburants (pages 34 et 35)).
Current law, horizon, limitations and 9 sources

Current law. Standard rate of 20% since 1 January 2014 (about 65% of taxable uses); reduced rates of 10%, 5.5% and 2.1%. Net yield 2025: €209bn for general government as a whole, of which €99bn for the State. Total net VAT forecast at €215.4bn in 2026 (PLF). A fraction also funds public broadcasting (compensation of €3.9bn in 2026).

Horizon. Full year, in 2025 euros; for 2032, of the order of +15 to +20% (assumption).

Limitations. Unit costs from the DG Trésor (Trésor-Éco no. 371, September 2025, static, model based on the 2022 accounts aged to 2025): yield of one point gross (table 1) and net (table 2), by rate. Part of VAT is paid by public administrations themselves: the net figure deducts the resulting increase in public spending. Cost schedule rule: net at the centre and as lower bound, gross as upper bound (standard rate 7.5 and 8.9; 10%: 1.6 and 1.9; 5.5%: 2.0 and 2.4; 2.1%: 0.4 and 0.4, equal after the publication's rounding; all rates: 11.4 and 13.7, published totals, the sum of the rounded rates giving 11.5 and 13.6). EU law (VAT directive revised in 2022) also governs reduced rates on fossil fuels. Regressive effect: −0.7% of purchasing power for the first quintile and −0.4% for the last, per point. Targeted cuts (energy, essential products) are costed on the base of each product. VAT at 5.5% on energy: €11.3bn is the static costing of the Institut Montaigne (2024), on 2022 VAT revenue (€15.6bn at the 20% rate on petroleum products, gas and electricity). Its range of €9bn to €13.6bn adds a 20% variation in consumption and prices: it is not a static costing and is not counted. Since 1 August 2025, electricity standing charges have been taxed at 20%: the base concerned is broader than in 2022. According to the same source, a reduced rate on motor fuels is not permitted by the VAT directive (annex III); for gas and electricity, it requires the Commission's agreement. This option concerns VAT only; a cut in excise duties is costed under the “Taxes on fuel and energy” and “Electricity prices, European market” levers. VAT at 5.5% on energy: €11.3bn is the gross loss (Institut Montaigne, 2022 data); the DG Trésor note says nothing about energy: the centre applies the net-to-gross ratio published for the 20% rate, at which this base is taxed (7.5 / 8.9 = 0.843), i.e. €9.52bn (calculation; 9.40 with the ratio for all rates, 11.4 / 13.7). Montaigne points out that the VAT directive does not allow a reduced rate on motor fuels. Energy sources taken one by one (options “gas and electricity”, “motor fuels and electricity” and “motor fuels, electricity and gas”), editors' calculations: Montaigne details 2022 VAT at the 20% rate by product (petroleum products €10,478m, natural gas €1,282m, electricity €3,872m; table based on the SDES 2022 monetary energy balance); moving from 20% to 5.5% removes 72.5% of it (1 − 5.5 / 20), i.e. €7.6bn (petroleum products), €0.93bn (gas) and €2.81bn (electricity) gross, the sum of which gives back the €11.3bn; same net-to-gross ratio at the centre (0.843). Gas and electricity: €3.74bn gross, 3.15 net. Motor fuels: VAT on petroleum products minus that on households' heating oil (€4,478.6m of spending including VAT in 2022, series N139MMR of the long series of the SDES 2022 energy balance, i.e. about €746m of VAT at the 20% rate), i.e. €9,732m, and a gross loss of €7.06bn; with electricity, €9.86bn gross, 8.31 net. Motor fuels, electricity and gas, without heating oil: 7.06 + 2.81 + 0.93, i.e. €10.79bn gross, 9.09 net. The rest of the VAT on petroleum products other than households' motor fuels (about €0.4bn, heating oil and other products bought by purchasers who do not deduct it) remains counted with motor fuels: slight overestimate. Reduced rates (parameters “10%”, “5.5%” and “2.1%”): the yield of one point of each is published by the DG Trésor (net at the centre, gross as upper bound); this is a primary source that directly costs the point, with no coefficient to derive. Measured range: one point of change; beyond that (reduced rates merged into a single rate, for example), linear extrapolation, marked as such, which ignores behavioural effects, shifts in consumption between rates and the limits placed on reduced rates by the VAT directive. The “+1 point” options for a reduced rate are worth one step. End of VAT on fuel excise duty (option, 18 October 2026, editors' calculation): 20% of gross excise duty on motor fuels (€31.3bn in 2025, Voies et moyens of the PLF 2027) × share of road excise duty paid by households (63.5% in 2022: €18.4bn out of €29.0bn, SDES), i.e. €3.98bn at the centre and as upper bound; lower bound, same calculation on the 2026 forecast (€30.2bn), €3.8bn. This is the same household share as in the excise duty coefficient (“Taxes on fuel and energy” lever). VAT paid on excise duty by businesses, which is deductible, is not net revenue; that paid by public administrations and by purchasers who do not deduct VAT is not counted (slight underestimate). EU law (according to the government's answer published on 3 March 2020, the VAT directive requires excise duties to be included in the VAT base) is not assessed, as elsewhere in the cost schedule. Works other than energy renovation at the standard rate (option, 18 October 2026): cost of tax expenditure 730213 relative to the standard rate, €2,465m forecast for 2026 (centre and upper bound), €2,195m actual in 2024 (lower bound), static. The 10% rate on passenger transport is not a tax expenditure (Voies et moyens, volume II, p. 10): no published costing of a move to 5.5%. Takeaway food is not separated from on-site dining in tax expenditure 730221.

Taxes on fuel and energycostable, revenue
  • Excise duty on fuels (change). For 1 cent per litre of excise duty on all products subject to the fuel excise (road fuels, heating oil…), central value including VAT on the excise paid by households: a gain of €0.53bn per year (0.47 to 0.6) (coefficient calculated by the editors from Fipeco, Les taxes sur les carburants). Current value: 0 c€/l. Range measured by the source: -59.4 to 59.4 c€/l; beyond it, linear extrapolation is flagged. Coefficient proportional to the taxed volume (editors' calculation): arithmetic range from abolition of the excise duty on diesel (−59.40 c€/l) to its doubling. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 4 sources

Current law. Excise duty on energy products (formerly TICPE) in 2026, excluding regional surcharges: 59.40 c€/l for diesel and 68.29 c€/l for SP95-E5, itself subject to VAT. Carbon component frozen since 2018. 2025 yield: €31.3bn of gross excise on fuels, of which €11.9bn transferred to local authorities and €1.2bn to the AFITF; €16.3bn net for the State after €1.7bn of refunds (Voies et moyens, PLF 2027); €30.5bn in national accounts according to Fipeco. Tax expenditures on energy: €6.3bn to €7.9bn in 2024.

Horizon. Full year; base trending downwards (electrification): the yield per cent in 2032 will be lower than today.

Limitations. Coefficient calculated by the editors: €31.3bn of gross excise duty in 2025 (Voies et moyens, PLF 2027) divided by an average rate of about 63 c€/l (59.40 for diesel, 68.29 for SP95-E5), i.e. about 50 billion litres taxed and €0.50bn per cent excluding VAT, 0.47 net of refunds (€1.7bn). Excise duty is itself subject to VAT (20%): for a household, one cent of excise duty also carries 0.2 cent of VAT; the VAT paid by households on excise duty is real revenue (lost if excise duty falls), that paid by businesses is recoverable. Central value, €0.53bn per cent: 0.47 excluding VAT, plus 20% VAT on the share of excise duty paid by households, 63.5% of road excise duty in 2022 (€18.4bn out of €29.0bn; SDES, 2022 energy balance, long series N022MMR and N033MMR), i.e. 0.2 × 0.50 × 0.635 ≈ 0.06 (editors' calculation; until 18 October 2026, 0.58, with VAT counted on all litres). Upper bound, 0.60: VAT included on all litres, before refunds. Lower bound, 0.47: excluding VAT. This is not a published official figure: it ignores the rates specific to each product (heating oil, off-road diesel) and refunds by sector (hauliers, farmers). State excise duties on motor fuels fall in 2026 (prices of petroleum products) then rise again in 2027 according to the Voies et moyens. A cut of 10 c€/l is given by the linear parameter.

Business taxation

Corporate income taxcostable, revenue
  • Standard corporate income tax rate. For 1 point on the standard corporate income tax (IS) rate, static (€59.9bn of net IS in 2025 divided by 25): a gain of €2.4bn per year (2.34 to 2.9) (coefficient calculated by the editors from Fipeco, L'impôt sur les sociétés). Current value: 25%. Range measured by the source: 25% only; beyond it, linear extrapolation is flagged. Coefficient calculated by the editors from the yield at the current rate (25%): any other value is a linear extrapolation, flagged as such.
  • Corporate income tax rate for large companies only (category of the Economic Modernisation Act, about 300 taxpayers). For 1 point of corporate income tax rate on large companies only: 1 point of the standard rate (€2.4bn) × their share of gross corporate income tax (€23.5bn out of €71.2bn in 2021, i.e. 33%): a gain of €0.79bn per year (0.77 to 0.96) (coefficient calculated by the editors from Fipeco, L'impôt sur les sociétés, et de Insee Références, Les entreprises en France, édition 2023, fiche 3.6 « Impôt sur les sociétés » (données DGFiP 2021)). Current value: 25%. Range measured by the source: 25% only; beyond it, linear extrapolation is flagged. Coefficient calculated by the editors (coefficient of the standard rate × share of large companies in gross corporate income tax, DGFiP 2021): any value other than 25% is a linear extrapolation, flagged as such.
Current law, horizon, limitations and 4 sources

Current law. Standard rate of 25% since 2022; 15% on the first €42,500 for SMEs (turnover < €10m); social contribution of 3.3%. Exceptional surcharge (LFI 2025) of 20.6% of corporate income tax if turnover exceeds €1bn and 41.2% if it exceeds €3bn (effective rate of 30.1% or 35.3%), renewed in the LFI 2026. Yield: €59.9bn of net corporate income tax in 2025, €60.3bn forecast for 2026 and €58.5bn in 2027 (Voies et moyens, PLF 2027), excluding the exceptional contribution (it appears among other revenue); €67.6bn in 2025, including the €7.5bn surcharge, according to Fipeco.

Horizon. Full year; for 2032, to be indexed to taxable profit, which is highly cyclical.

Limitations. No official cost per point of corporate income tax (IS) was found: the impact assessment of the cut in the rate from 33⅓% to 25% (PLF 2018) is accessible neither in the National Assembly documents nor in the Senate general report consulted. Coefficient calculated by the editors: central value, €59.9bn of net IS in 2025 divided by 25; lower bound, €58.5bn forecast for 2027 divided by 25, i.e. 2.34; upper bound, 1% of the €287bn of profits of profit-making companies, before loss carry-forwards (Fipeco). High elasticity (profit shifting): a static costing overstates the gain from an increase and the cost of a cut. Rate for large companies only (rule on public bases, 18 October 2026): 1 point is worth the standard-rate coefficient multiplied by the share of large companies (300 taxpayers) in gross IS, €23.5bn out of €71.2bn in 2021 (DGFiP, published by Insee), i.e. 33%; their share in IS net of tax credits is similar (€19.4bn out of €60.0bn, 32%). In 2021, companies with turnover above €250m were taxed at 27.5% and the others at 26.5%: the 2021 share roughly reflects that of profits. The exceptional surcharge on large companies (“Windfall profits and share buybacks” lever) is renewed for a third and final financial year by article 15 of the finance bill for 2027, at reduced rates: it is not in the law in 2032.

  • Fipeco, L'impôt sur les sociétés, 2026-09-30: « En 2025, le produit de l'IS était de 67,6 Md€ en comptabilité budgétaire ... majoration exceptionnelle incluse ... Son rendement a été de 7,5 Md€ en 2025 ... 1,8 million d'entreprises ... bénéficiaires (287 Md€ de bénéfices) »
  • PLF 2027, Évaluation des voies et moyens, tome I (page 22), 2026-10: « Les recettes d’impôt sur les sociétés net ont atteint 59,9 Md€ en 2025, soit une plus-value de +6,9 Md€ par rapport à la prévision de la LFI 2025. […] Les recettes d’impôt sur les sociétés sont estimées à 60,3 Md€ en 2026. […] En 2027, les recettes d’impôt net sur les sociétés s’élèveraient à 58,5 Md€ »
  • Insee Références, Les entreprises en France, édition 2023, fiche 3.6 « Impôt sur les sociétés » (données DGFiP 2021), 2023: « L’IS brut des grandes entreprises, d’un montant de 23,5 milliards d’euros en 2021 […] l’IS brut, c’est‑à‑dire l’impôt dû avant prise en compte des réductions et crédits d’impôt, s’élève à 71,2 milliards d’euros […] Les 300 grandes entreprises bénéficient de 4,1 milliards d’euros de crédit d’impôt […] Le montant des crédits d’impôt perçu par les entreprises à l’IS a quant à lui légèrement diminué entre 2020 et 2021 (11,2 contre 11,6 milliards d’euros) […] le taux d’IS est porté à 27,5 % pour les entreprises réalisant plus de 250 millions d’euros de chiffre d’affaires et à 26,5 % pour les entreprises réalisant jusqu’à 250 millions d’euros de chiffre d’affaires. »
  • Legifiscal, PLF 2027 : prorogation de la contribution exceptionnelle sur les bénéfices des grandes entreprises à taux plus bas (1er octobre 2026), 2026-10-01: « L'article 15 l'étend à un troisième exercice, soit l'exercice 2027 pour une clôture au 31 décembre. […] Le taux passerait de 20,6 % à 15,7 % pour un chiffre d'affaires compris entre 1,5 et 3 milliards d'euros, et de 41,2 % à 31,4 % au-delà de 3 milliards d'euros. […] la contribution, réduite de 30 %, rapporterait 5 milliards d'euros en 2027, contre 7,5 milliards en 2026. »
Taxation of digital giants and multinationalscostable, revenue
  • Digital services tax rate. For 1 point on the digital services tax rate (€981m forecast for 2027 at the 3% rate, divided by 3): a gain of €0.327bn per year (0.302 to 0.327) (coefficient calculated by the editors from PLF 2027, Évaluation des voies et moyens, tome I (pages 24, 25, 29 et 30)). Current value: 3%. Range measured by the source: 3% only; beyond it, linear extrapolation is flagged. Coefficient calculated by the editors from the yield at the observed rate (3%): any other value is a linear extrapolation, flagged as such, which ignores behavioural and threshold effects.
  • Spending funded by the tax revenue (earmarked revenue). For €1bn of tax revenue allocated, in the same sentence, to a named spending item: the spending is counted at the same amount as the revenue (earmarked revenue rule): a cost of €1bn per year. Current value: 0 €bn/year.
  • Abolition of the digital services tax: a cost of €0.981bn per year (0.906 to 0.981).
Current law, horizon, limitations and 3 sources

Current law. Digital services tax (created in 2019, code des impositions sur les biens et services, Art. L. 453-45 et seq.): 3% of revenue earned in France from digital intermediation, targeted advertising and the sale of data, for companies whose worldwide turnover exceeds €750m and taxable revenue in France €25m; €902m in 2025, €906m forecast for 2026, €981m in the PLF 2027. 15% global minimum tax (Pillar 2, Directive (EU) 2022/2523, transposed from the Finance Act for 2024): in force for 2024 financial years, first collections in 2026 (€500m forecast), no yield forecast for 2027. Since 5 January 2026, the OECD “side-by-side” regime exempts groups that opt for it, starting with groups with a US parent company, from the IIR and UTPR rules, for financial years beginning on or after 1 January 2026.

Horizon. Full year; cost = current yield of the tax (static), PLF 2027 forecast as central value, revised forecast for 2026 as lower bound.

Limitations. Earmarked revenue: when the position, in the same sentence, earmarks the tax revenue for a named expenditure, that expenditure is counted at the same amount as the costed revenue (central value), via the parameter “spending financed by the tax revenue”; the revenue and the spending each go into their own total. Abolishing the digital services tax is costed at its revenue. The rate parameter applies the average yield of one point at the current rate (coefficient calculated by the editors: €981m / 3; €906m / 3 as lower bound): an increase to 6% or 15% (proposed by amendment in the National Assembly in autumn 2025 without a revenue assessment; the finance act for 2026 kept 3%) is a linear extrapolation, marked as such, that no administrative source (Voies et moyens, impact assessment, general report) costs. Nor does any source cost the adjustment of multinationals' profits in proportion to the turnover achieved in France (« taxation unitaire », unitary taxation): only estimates by elected officials exist, not counted. The global minimum tax has no forecast yield in 2027: according to the Voies et moyens, countries have changed their tax rules and the exceptional contribution of large companies raises their effective rate; a commitment to apply or strengthen it therefore has no yield that can be costed from this source. The tax's yield varies with the platforms' turnover in France; it is not updated to 2032.

  • PLF 2027, Évaluation des voies et moyens, tome I (pages 24, 25, 29 et 30), 2026-10: « Elle s’applique à certaines entreprises du secteur numérique dont le chiffre d’affaires mondial dépasse 750 M€ et dont les recettes taxables en France excèdent 25 M€. […] Son taux est fixé à 3 % et porte sur les recettes générées par ces services en France. […] Les recettes se sont élevées à 902 M€ en 2025 […] En 2026, la prévision s’établit à 906 M€ […] La prévision s’établit à 981 M€ pour 2027, soit une hausse de +75 M€ par rapport au révisé 2026. […] L’impôt minimum mondial est entré en vigueur pour les exercices 2024 […] La prévision révisée pour 2026 s’établit à 500 M€ […] Pour 2027, aucun rendement n’est prévu au titre de l’impôt minimum mondial. »
  • OCDE, Cadre inclusif sur le BEPS, « Side-by-Side Package » (pages 2, 8 et 84), 2026-01-05: « This document was approved and declassified by the OECD/G20 Inclusive Framework on BEPS on 5 January 2026 […] When it elects for the safe harbour, an MNE Group will not be subject to the IIR or UTPR. […] The SbS Safe Harbour is applicable for Fiscal Years commencing on or after 1 January 2026 »
  • Déclaration du G7 sur l'imposition minimale mondiale, 28 juin 2025 (publiée par le ministère des Finances du Japon), 2025-06-28: « A side-by-side system would fully exclude U.S. parented groups from the UTPR and the IIR in respect of both their domestic and foreign profits. »
Production taxescostable, revenue
  • Abolition of the CVAE (business value-added contribution) (already scheduled for 2030: zero cost in 2032 compared with current policies): a cost of €0bn per year.
  • Retention or reinstatement of the CVAE (business value-added contribution) beyond 2030 (revenue compared with current policies): a gain of €4.11bn per year (4.02 to 4.11).
  • Abolition of the C3S (company social solidarity contribution): a cost of €5.2bn per year.
Current law, horizon, limitations and 4 sources

Current law. Taxes on production: €132.8bn in 2025. Including CVAE: €4,024m of State revenue in 2025, €4,106m forecast for 2026 and €4,110m in 2027; its abolition, planned for 2027, was postponed to 2030 by the Finance Act for 2025 (Voies et moyens, PLF 2027; €3.7bn paid by businesses in 2025 according to Fipeco). The Finance Act for 2026 contains no article on the CVAE. C3S: €5.2bn (2024). CFE: €7.7bn. Property taxes paid by businesses are included in the €43.1bn of TFPB.

Horizon. Full year; cost to the State, which compensates local authorities with a fraction of VAT.

Limitations. Costs = current revenue (static). The abolition of the CVAE (value-added contribution of businesses) in 2030 is confirmed by the Voies et moyens of the PLF 2027 (October 2026); the text of the code général des impôts (General Tax Code) in force could not be consulted on Légifrance (automated access refused), and the adopted text of the finance act for 2026 does not change this timetable. Compared with current policies, the cost in 2032 of an abolition is therefore zero, and maintaining it is a gain: €4.02bn (2025, lower bound) to €4.11bn (2027, central value and upper bound). Cut announced as an amount without a list of the taxes targeted: a pure amount with no named instrument, not costed, cited next to the total (decision 1 of 18 October 2026, the same for all candidates); the reduction in aid announced in return is treated in the same way (“Public aid to businesses” lever). The choice of taxes does not change the static cost; an amount greater than the revenue of a named tax cannot come from that tax alone (CVAE €4.1bn, C3S 5.2, CFE 7.7; €132.8bn for all production taxes in 2025).

  • Fipeco, La CVAE, 2026-07-07: « La loi de finances initiale pour 2025 a majoré le taux de la CVAE applicable en 2026 et 2027 et reporté sa suppression en 2030 ... En 2025, la CVAE prélevée sur les entreprises s'est élevée à 3,7 Md€ ... Le produit de l'ensemble des impôts sur la production s'est élevé à 132,8 Md€ en 2025 »
  • Fipeco, Les impôts sur la production, 2025-11-12: « contribution sociale de solidarité des sociétés, pour 5,2 Md€ ... la cotisation foncière des entreprises (7,7 Md€) »
  • OFCE, Le déficit public devrait diminuer en 2026 et 2027, 2026-04: « Les entreprises seraient plus mises à contribution en 2026 (+7,4 milliards d’euros, 0,2 point de PIB), en raison de l’annulation de la suppression progressive de la CVAE (1,3 milliard d’euros,) mais surtout de l’augmentation de la contribution exceptionnelle sur les grandes entreprises (7,3 milliards, 0,2 point de PIB). »
  • PLF 2027, Évaluation des voies et moyens, tome I (page 32), 2026-10: « Prévue pour être supprimée en 2027, la CVAE sera finalement maintenue jusqu’en 2030, conformément à la loi de finances pour 2025. RETOUR SUR 2025 En 2025, les recettes de CVAE pour l’État ont atteint 4 024 M€, en moins-value de -6 M€ par rapport à la prévision de la LFI pour 2025. […] En 2026, la prévision s’établit à 4 106 M€ […] La prévision s’établit à 4 110 M€ en 2027 »
Windfall profits and share buybackscostable, revenue
  • Make permanent the corporate income tax (IS) surcharge on large companies at the 2025 rate: a gain of €7.5bn per year (7.3 to 8).
  • Make permanent at half rate (PLF 2026): a gain of €4bn per year (4 to 6).
  • Tax on capital reductions through share buybacks raised from 8% to 33%, on the current base (permanent yield of €200m at 8%, Voies et moyens), gain net of the current yield, static (editors' calculation): a gain of €0.63bn per year (coefficient calculated by the editors from PLF 2026, Voies et moyens tome I).
  • 4% tax on dividends paid by companies (2% below €10m distributed), static: yield of the 3% contribution on distributed amounts (2013-2017) raised to 4% (editors' calculation): a gain of €2.4bn per year (coefficient calculated by the editors from Assemblée nationale, question au Gouvernement n° 203 (Charles de Courson), 2017).
Current law, horizon, limitations and 4 sources

Current law. Exceptional contribution on the profits of large companies (turnover > €1bn): €8bn forecast for 2025 and €7.5bn collected. Renewed in the LFI 2026 at a reduced level: €4bn in the PLF, €6bn according to the Senate, €7.3bn according to the OFCE. Tax on share buybacks (LFI 2025): €0.4bn in 2025, €0.2bn per year on a permanent basis.

Horizon. Full year; the surcharge is not in the law beyond 2026: making it permanent is a gain compared with current policies.

Limitations. 2026 yield of the renewal: three sources diverge (€4bn / €6bn / €7.3bn). A « taxe sur les superprofits » (superprofits tax) on excess profit relative to an average requires a precise definition. No generic official costing. The tax on share buybacks is already in force. Tax on share buybacks at 33% (option, editors' calculation, 18 October 2026): the tax in force (8%, consolidated turnover above €1bn) raises €200m per year at steady state; at 33% on the same base, 0.2 × 33 / 8 = €0.83bn, i.e. €0.63bn more. A 33% rate would discourage buybacks: static costing, upper limit; extending the base to the acquisition value of the shares and to companies with turnover above €750m has no published base. 4% tax on dividends paid (option, editors' calculation, 18 October 2026): the 3% contribution on amounts distributed « a rapporté à l'État 1,8 milliard d'euros par an entre 2013 et 2017 » [“brought the State 1.8 billion euros per year between 2013 and 2017”] (question to the Government by Charles de Courson, National Assembly); at 4%, 1.8 × 4 / 3 = €2.4bn. Two biases in opposite directions: companies distributing less than €10m would pay 2% (counted as less), while SMEs exempt from the former contribution and the rise in dividends since 2017 are not counted (more). The 3% scheme was struck down (Conseil constitutionnel, 6 October 2017) and refunded (about €10bn); the legal compliance of a new tax is not assessed, as elsewhere in the cost schedule.

Financial transaction taxcostable, revenue
  • Financial transaction tax rate. For +0.1 point on the financial transaction tax rate (Senate finance committee: €600m for moving from 0.3% to 0.4%): a gain of €0.6bn per year. Current value: 0.4%. Range measured by the source: 0.3% to 0.4%; beyond it, linear extrapolation is flagged. Range measured by the source: from 0.3% to 0.4% (increase costed by the Senate finance committee). Beyond that, linear extrapolation, flagged as such, which ignores the response of transaction volumes.
  • Abolition of the financial transactions tax: a cost of €2.973bn per year (2.888 to 2.973).
Current law, horizon, limitations and 3 sources

Current law. Financial transaction tax (General Tax Code, Art. 235 ter ZD): 0.4% of the amount of acquisitions of shares in French companies with a market capitalisation above €1bn (rate raised from 0.3% to 0.4% by the Finance Act for 2025); bonds and derivatives are not taxed. Yield: €2,526m in 2025, €2,888m forecast for 2026, €2,973m in the PLF 2027.

Horizon. Full year; abolition: current yield of the tax (static), PLF 2027 forecast as central value, revised forecast for 2026 as lower bound; rate: yield of an increase costed by the Senate in December 2024, for 2025.

Limitations. Abolition is costed at its revenue. The rate parameter applies the only institutional costing of an increase read, that of the Senate finance committee for the move from 0.3% to 0.4% (€600m, in force since 2025), a primary source that replaces the coefficient previously calculated by the editors (average yield of 0.1 point at the current rate: €2,973m / 4 = €0.74bn, higher, not used); any value above 0.4% is a linear extrapolation, marked as such, which ignores the reaction of the volume of transactions. No administrative source costs an extension of the base (intraday transactions, derivatives). Authors' costing, not counted: a move to 0.5% would bring in about €1.5bn more according to its authors (Senate, amendment I-950 to the PLF 2025, calculated on the revenue forecast for 2023). The yield follows the volume and price of transactions; it is not updated to 2032.

  • PLF 2027, Évaluation des voies et moyens, tome I (pages 45 et 46), 2026-10: « La taxe sur les transactions financières s’applique aux acquisitions d’actions émises par les sociétés françaises dont la capitalisation boursière dépasse 1 milliard d’euros. Depuis la loi de finances pour 2025, son taux a été relevé à 0,4 % du montant total de la transaction. […] Les obligations et les produits dérivés ne sont pas soumis à la taxe. […] Les recettes se sont élevées à 2 526 M€ en 2025 […] La prévision s’établit à 2 888 M€ en 2026 […] La prévision s’établit à 2 973 M€ en 2027, soit une hausse de +85 M€ par rapport à 2026. »
  • Sénat, PLF 2025, amendement n° A-24 de la commission des finances (article 26 quater, adopté), 2024-12-01: « La recette supplémentaire résultant du passage de 0,3 % à 0,4 % du taux de la taxe est estimée à 600 millions d'euros. »
  • Sénat, PLF 2025, amendement n° I-950 (estimation de ses auteurs), 2024-11-20: « une telle augmentation permettrait à l'État de collecter environ 3,73 milliards d'euros, contre 2,24 milliards d'euros actuellement avec un taux de 0,3 %, soit une hausse annuelle de 1,5 milliard d'euros »

Debt and public spending

Public deficit targeted for 2032not costable
Current law, horizon, limitations and 3 sources

Current law. Public deficit 2025: 5.1% of GDP (€152.5bn). 2026 target: 5.0%, maintained in the RAA 2026. Official path: « sous 3 % du PIB à horizon 2029 » [“below 3% of GDP by 2029”] (PSMT 2025-2029, adjustment spread over 7 years with the agreement of the Council of the EU). OFCE forecasts: 4.8% in 2026, 4.4% in 2027. No official target published for 2032.

Horizon. Level in 2032, as % of GDP, to be compared with the extended PSMT (medium-term fiscal-structural plan) path (below 3%).

Limitations. Balance target, not a measure: it serves to compare the announced target with the total of costed measures. 1 point of GDP is worth about €36bn in 2032 (nominal growth assumption). It is not a cost lever as such: it is the targeted balance, which serves to test the consistency between costed measures and the announced target. The conversion into €bn in 2032 rests on a nominal GDP assumption: about €3,000bn in 2025 rising to €3,560bn to €3,680bn.

Level of public spendingnot costable
Current law, horizon, limitations and 2 sources

Current law. Public spending 2025: €1,714bn, i.e. 57.2% of GDP (57.0% in 2024); euro area 49.8%, Germany 50.5%. European norm: net primary expenditure capped at +1.2% in 2026, 2027 and 2028 and +1.1% in 2029 (Council recommendation of 21 January 2025).

Horizon. Ratio in 2032; gap with the PSMT path.

Limitations. Targeted level of spending, not a measure: consistency check against the sum of costed measures. Outcome lever, not a measure lever: it serves as a consistency check against the sum of measures. Fipeco: « l'écart de 6,7 points avec l'Allemagne = 200 Md€ » [“the gap of 6.7 points with Germany = €200bn”].

Announced savingscostable, spending
  • Saving on the State grant to the Presidency of the Republic. For €1m saving on the grant to the Presidency of the Republic: a gain of €0.001bn per year (candidate’s amount, taken as is). Current value: 0 €m.
  • Research tax credit reserved for SMEs and mid-sized companies: the claims of large companies abolished (42% of claims in 2019): a gain of €3.24bn per year (3 to 3.41) (coefficient calculated by the editors from MESR, « Le crédit d'impôt recherche en 2019 » (données semi-définitives)).
Current law, horizon, limitations and 4 sources

Current law. LFI 2026 (OFCE calculations): structural primary effort by the State, excluding the EU contribution, of €11bn (0.4 points of GDP), €16bn on its other items when the increase in the defence budget is taken into account; savings of €4bn in the LFSS and €3bn for local authorities (0.1 points of GDP each); overall budgetary effort of €16bn (0.5 points of GDP), €22bn excluding the increase in interest charges. In 2025, net primary expenditure grew by only €12.5bn (+0.8%), thanks to €23bn of new revenue.

Horizon. Savings in 2032 compared with the baseline trend; the baseline itself is not published (Fipeco).

Limitations. An announced amount of savings is counted only through the measures that make it up; on its own, it is a target. Benchmark: primary savings in the 2026 initial finance act (LFI), €18bn. An announced amount is counted only if it is backed by identified measures; the rest is a target, flagged and not added. Bercy's baseline trend is not published, which makes « économies par rapport au tendanciel » [“savings relative to the baseline trend”] unverifiable. A global amount overlaps the sectoral levers. Appropriation for the Presidency: amount in the PLF 2026 (€122.56m, stable since 2025); that of the finance act as voted is not cited. One euro less of appropriation is worth one euro of savings for the State; the Élysée's spending budget (€126.28m in 2026) is slightly broader than the appropriation. Option cir-reserve-pme-eti (rule on public bases, 18 October 2026: a named saving is calculated on its measure): in 2019, large companies received €3bn of research tax credit, 45% of the « CIR recherche » [“research CIR”] and 42% of the total claim (€7.12bn, MESR (Ministry of Higher Education and Research)); applied to the cost of the CIR in 2026 (€7.7bn to €8.1bn, Voies et moyens, option fin-cir of the “Tax and social security expenditures” lever), this share gives €3.24bn to €3.41bn; lower bound, the 2019 amount (editors' calculation; revenue loss assuming unchanged behaviour, like any tax expenditure). The CIR is a tax expenditure: declared overlap with its abolition and with the across-the-board cut in tax expenditures.

  • OFCE, Le déficit public devrait diminuer en 2026 et 2027, 2026-04: « les économies attendues issues la Loi de finances de la Sécurité Sociale (LFSS 2026) […] atteindraient 4 milliards d’euros en 2026 (0,1 point de PIB). […] Les économies prévues du côté des collectivités locales seront de 3 milliards d’euros en 2026 (0,1 point de PIB). L’État devrait ainsi réaliser un effort structurel primaire hors contribution à l’UE (PSR-UE) de 11 milliards en 2026 (0,4 point de PIB). Si l’on tient compte de la hausse substantielle du budget de la défense en 2026 (+0,2 point de PIB), alors l’effort de l’État sur les autres postes budgétaires devra être de 16 milliards d’euros (0,5 point de PIB). […] Cela représente un effort budgétaire global de 16 milliards d’euros en 2026 (0,5 point de PIB) et de 22 milliards d’euros si l’on exclut la hausse de la charge d’intérêts (soit 0,7 point de PIB). »
  • Fipeco, La croissance tendancielle des dépenses publiques, 2026-09-18: « la direction générale du trésor ... et la direction du budget établissent des scénarios tendanciels à moyen terme ... Leurs travaux ne sont pas publiés. »
  • Sénat, commission des finances, « L'essentiel » sur la mission Pouvoirs publics, PLF 2026 (rapporteur spécial Grégory Blanc), 2025-10: « « Le budget 2026 prévoit des dépenses de 126,28 millions d'euros, soit une hausse de 0,54 % (+ 0,68 millions d'euros) par rapport à 2025, dans un contexte de stabilité de la dotation de l'État (à 122,56 millions d'euros) » ; tableau : « 501 - Présidence de la République 122 563 852 » »
  • MESR, « Le crédit d'impôt recherche en 2019 » (données semi-définitives, octobre 2022), 2022-10: « En 2019, le montant du crédit d'impôt recherche (CIR) s'élève à 7,12 Md€ pour les trois dispositifs, dont 95 % au titre de la recherche […] Au titre du seul « CIR recherche », les PME forment 83 % des 15 700 bénéficiaires et apportent 26 % du total des dépenses de recherche déclarées, soit une créance de 1,9 Md€ (28 %, un niveau similaire à celui des entreprises de taille intermédiaire, ETI). Les grandes entreprises (GE) bénéficiaires déclarent 48 % des dépenses de recherche et bénéficient d'une créance de 3 Md€ (45 % de la créance au titre de la recherche). »
Handling of the debtnot costable
Current law, horizon, limitations and 3 sources

Current law. Public debt (Maastricht) at end-2025: €3,460.5bn, i.e. 115.6% of GDP. OFCE forecasts: 118.3% in 2026, 119.8% in 2027. General government interest charges: €64.7bn in 2025, i.e. 2.2% of GDP (OFCE: 2.4% in 2026, 2.6% in 2027). Average maturity of State debt: 8.5 years (April 2026).

Horizon. Not applicable in static costing; sensitivity given as a benchmark.

Limitations. The « traitement » (treatment) of the debt (cancellation of securities held by the Banque de France, restructuring, ring-fencing, forced loan) is a matter of European law and market confidence: it is not costed statically. Sensitivity benchmark: +100 basis points across the whole yield curve add €3.1bn of interest charges in the 1st year, €7.5bn in the 2nd, €18.4bn in the 5th, €32.1bn in the 10th (negotiable State debt); +1 point of inflation adds €3.0bn (index-linked securities). To be used to assess the effect of a programme on the risk premium, separately from the central figure.

Fiscal rulesnot costable
Current law, horizon, limitations and 2 sources

Current law. Stability Pact revised in April 2024 (EU Regulations 2024/1263 and 2024/1264): reference values of 3% for the deficit and 60% for debt; operational criterion = net primary expenditure path of the PSMT 2025-2029. For France, under the excessive deficit procedure since July 2024: reduction of the structural deficit by at least 0.5 points of GDP per year as long as the deficit exceeds 3%. Possible sanction of up to 0.05% of GDP every six months (about €1.5bn). In domestic law: LOLF, LPFP 2023-2027, HCFP.

Horizon. Not applicable.

Limitations. Institutional lever: exit from or renegotiation of the rules, golden rule, exclusion of investment or defence (national escape clause). No static cost; encoded as direction of change. The escape clause for defence allows a limited deviation.

  • Sénat, L'essentiel sur le RAA 2026, 2026-04: « Elle fait de l'évolution des dépenses primaires nettes ... le principal critère opérationnel ... une sanction financière pouvant aller jusqu'à 0,05 % de son PIB de l'année précédente, soit environ 1,5 milliard d'euros, tous les six mois »
  • Fipeco, Les nouvelles règles budgétaires européennes, 2026-04-14: « cette trajectoire doit permettre de réduire le déficit structurel de 0,5 point de PIB par an tant que le déficit public est supérieur à 3,0 % du PIB »

Local authorities

Local taxationcostable, revenue
  • Reinstate a TH (housing tax) on main residences (yield before abolition, 2017 euros, not updated): a gain of €22.2bn per year (19.3 to 23.2).
  • 10% increase in the yield of DMTO (property transfer duties) (derived): a gain of €1.8bn per year (coefficient calculated by the editors from Fipeco, Les impôts locaux).
  • VAT share of €9,000 paid by the State to the municipality for each new home started (274,611 in 2025, SDES), counted as a new grant (editors' calculation): a cost of €2.47bn per year (coefficient calculated by the editors from SDES, Construction de logements : résultats à fin décembre 2025 (France entière)).
Current law, horizon, limitations and 3 sources

Current law. Compulsory levies of local government: €190bn in 2025. Residence tax (taxe d'habitation) on main residences abolished since 2023, offset by the transfer of the départements' TFPB to municipalities and by fractions of VAT. Remaining: TH on second homes (€3.1bn, rate set freely); TFPB (€43.1bn); CFE; DMTO (€17.7bn in 2025); transferred fraction of VAT (€53bn).

Horizon. Local revenue; impact on the overall general government balance, neutralised if the measure merely transfers between the State and local authorities.

Limitations. The TH (housing tax) figures date from 2017 to 2021 (€22.2bn received by local authorities; €19.3bn paid by households) and are not updated. Some local reforms are merely transfers, with a zero balance for general government as a whole: treated as such. Exception (option of 18 October 2026): a State payment to communes earmarked for new spending (public facilities) is counted as a grant, at the cost to the State, as in the local authority grants lever: €9,000 × 274,611 housing starts in 2025 (SDES, seasonally adjusted cumulative total) = €2.47bn per year, at the 2025 construction rate (editors' calculation; 379,222 dwellings authorised in the same year).

  • Fipeco, Les impôts locaux, 2026-06-24: « Les prélèvements obligatoires affectés aux administrations publiques locales se sont élevées ... à 190 Md€ en 2025 ... Son produit s'est élevé à 22,2 Md€ pour les collectivités locales en 2017 ... les ménages n'ont payé que 19,3 Md€ ... Les droits de mutation à titre onéreux (DMTO) ont rapporté 17,7 Md€ en 2025 »
  • Fipeco, Les recettes des administrations publiques locales, 2026-07-06: « la TVA transférée (53 Md€) »
  • SDES, Construction de logements : résultats à fin décembre 2025 (France entière), 2026-01: « Sur toute l'année 2025, 379 222 logements ont été autorisés à la construction, soit 15,0 % de plus que sur l'année 2024 mais 8,8 % de moins que la moyenne des cinq années précédentes. […] Sur toute l'année 2025, 274 611 logements ont été mis en chantier, soit 21,3 % de moins que la moyenne des cinq années précédentes. »
State grants to local authoritiescostable, spending
  • Increase in State financial transfers to local authorities. For 1% increase in State financial transfers to local authorities: a cost of €0.49bn per year (0.49 to 0.53) (coefficient calculated by the editors from Fipeco, Les recettes des administrations publiques locales). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • Additional annual envelope paid to local authorities. For +€1bn a year paid to local authorities: a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
  • Cancel the 2026 effort required of local authorities: a cost of €3bn per year.
  • Communes' DGF (general operating grant) (€12.8bn in 2025) indexed to inflation each year (editors' common assumption: 1.75% per year, 1.5 to 2.0%; five upratings from 2028 to 2032, i.e. a cumulative increase of 9.1% in 2032), compared with a DGF stable in current euros; reading key: a one-point gap for one year costs €0.13bn: a cost of €1.16bn per year (0.99 to 1.33) (coefficient calculated by the editors from réponse du Gouvernement à la question orale n° 545 (Assemblée nationale)).
Current law, horizon, limitations and 4 sources

Current law. State financial transfers: €49bn in the LFI 2025 (€45bn of deductions from revenue and €4bn of appropriations from the RCT mission). DGF: €27bn in 2025, compared with €40bn in 2014. In 2026, the regions' DGF is brought back into the budget (−€5.2bn of transferred VAT) and the effort asked of local authorities is about €3bn. The LPFP 2023-2027 provided for transfers of €53bn in 2023 and €56bn in 2027.

Horizon. Full year; to be expressed relative to the extended LPFP (public finance programming act) path.

Limitations. An envelope announced in euros (for example €10bn over five years) is costed through the parameter in €bn per year, €1 announced = €1 of State spending, not as a percentage of State transfers to local authorities. An increase in grants worsens the State balance but not necessarily that of general government as a whole, if local authorities save or invest. The scope (with or without transferred VAT) changes the amounts: €49bn to €56bn. Indexation of the communes' DGF (general operating grant) to inflation (option, rule on public bases of 18 October 2026): same common assumption as for the public sector pay index point and pensions, inflation of 1.75% per year at the centre (Cour des comptes), 1.5 to 2.0% for the range, first effect on 1 January 2028, cost in 2032 = cumulative gap of five upratings with the current rule (a DGF set each year in the finance act, which remained at €27.4bn in 2025 and in 2026). Communes' DGF in 2025: €12.8bn (Government); 12.8 × (1.0175^5 − 1) = €1.16bn (12.8 × 7.73% = 0.99; 12.8 × 10.41% = 1.33; editors' calculations, 2025 euros). Reading key: one point of indexation for one year is worth €0.128bn for the communes, €0.27bn for the whole DGF. Inter-municipal bodies, departments and regions are not counted. The DILICO (cyclical smoothing mechanism for local authority revenue: €740m in 2026, paid by the regions, the departments and the inter-municipal bodies, 90% paid back to the contributors in thirds over three years, 10% to the horizontal equalisation funds) places local revenue in reserve and returns it to local authorities: as the whole levy goes back to local authorities, abolishing it costs the State nothing in 2032: if renewed each year, that year's levy is returned in 2032 through the repayments of previous years; if not renewed, it no longer exists (editors' reading).

Public services

School

Pupils per class, primarycostable, spending
  • Average pupils per class, state primary schools. For 1 pupil fewer per class on average, state primary schools (about 11,800 additional teaching posts, at 2029-2032 enrolment): a cost of €0.5bn per year (0.5 to 0.85) (coefficient calculated by the editors from Sénat, rapport spécial PLF 2026, Enseignement scolaire (O. Paccaud)). Current value: 21 pupils. Range measured by the source: 20 to 21 pupils; beyond it, linear extrapolation is flagged. The cost is not linear (each pupil fewer costs more than the previous one): beyond one pupil fewer, linear extrapolation, flagged as such, underestimates it.
  • Average reduced to 19 pupils per class, public primary schools (about 24,800 additional teaching posts, at 2029-2032 pupil numbers): a cost of €1.05bn per year (1.05 to 1.74) (coefficient calculated by the editors from Sénat, rapport spécial PLF 2026, Enseignement scolaire (O. Paccaud)).
Current law, horizon, limitations and 3 sources

Current law. No general legal cap. Cap of 24 pupils in grande section (final year of nursery school), CP and CE1, and halving of these classes in REP/REP+ since 2017. Average in public elementary schools: 21.3 pupils per class in 2024; the minister announces 21.0 for the first time at the start of the 2025 school year. Primary level (public and private): 6.15 million pupils today, fewer than 5.7 million forecast in 2029 (DEPP, cited by the Senate).

Horizon. full year 2032, €bn/year, compared with current policies

Limitations. Calculation: posts = P/(s×(s−1)), with P ≈ 4.95m public-sector pupils in 2029-2032 (editors' assumption: 87% of 5.7m) and s = 21, i.e. about 11,800 posts. Cost per post at the centre: €48,336 excluding the CAS Pensions (State pensions special account), brought back to pay for active service (48,336 / 1.145 ≈ €42,200), i.e. €0.5bn; high value: about €70k per post counting the contribution to the CAS Pensions. “Net” convention (R9), the same as for the public sector pay index point and public sector headcount: at the centre, cost excluding all employer contributions paid to other public administrations (CAS Pensions, employer social contributions); as upper bound, full employer cost. The published “excluding CAS” cost still includes employer contributions other than the CAS: it is brought back to pay for active service alone by the ratio “average loaded cost excluding CAS / pay for active service” from the PAP 2024 (Enseignement scolaire, entry-exit costs: 47,634 / 41,610 = 1.145 in primary, 51,198 / 44,265 = 1.157 in secondary, about 1.15 on average; editors' calculation). The cost is not linear: each pupil fewer costs more than the previous one (going from 21 to 18 costs more than three times the unit). Decisive point: current policies cut posts in line with demographics (4,018 in 2026). A candidate who “maintains posts” therefore reduces class sizes at no cost compared with 2026, but at a cost compared with the reference trajectory. The DEPP note on class sizes and the RERS (education.gouv.fr) were not accessible: the figures come from the Senate, not from the DEPP. Option moyenne-19 (average of 19): same method, same sources and same assumptions as the parameter (4.95m public-sector pupils in 2029-2032). Posts = P/19 − P/21 = 4.95m × (1/19 − 1/21) ≈ 24,800, at about €42,200 (€48,336 excluding the CAS Pensions, brought back to pay for active service), i.e. €1.05bn; €1.74bn at about €70k per post with the contribution to the CAS Pensions. Caution: the lever is defined as a maximum number of pupils per class, whereas the cost schedule costs an average. A cap at a given level costs considerably more than an average at the same level. A position is therefore costed only if the text targets an average.

Teachers' paycostable, spending
  • Increase in pay for all teachers. For +1% in pay for all teachers (state and private under contract): a cost of €0.357bn per year (0.357 to 0.59) (coefficient calculated by the editors from Sénat, rapport spécial PLF 2026, Enseignement scolaire). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • Uniform increase in net pay for all teachers, in euros per month. For +€100 net a month for each of the 852,752 paid teachers (state and private under contract, 2024): a cost of €1.19bn per year (1.16 to 1.91) (coefficient calculated by the editors from Sénat, rapport spécial PLF 2026, Enseignement scolaire). Current value: 0 € net per month. Range measured by the source: 0 to 3,090 € net per month; beyond it, linear extrapolation is flagged. Coefficient proportional to the number of teachers (editors' calculation): arithmetic range up to doubling the average net salary of a tenured teacher (€3,090 a month in 2024). Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 4 sources

Current law. Civil service pay scales, public sector pay index point frozen at €4.9228 since 1 July 2023. 2023 « socle » (baseline) pay rise (about +10% on average, mainly at the start of careers) and the Pacte enseignant (teacher pact). In 2024, 712,806 public-sector teachers and 139,946 in state-contracted private schools are paid. Title 2 of the Enseignement scolaire (school education) mission: €59.23bn excluding the CAS Pensions (PLF 2026).

Horizon. full year 2032, €bn/year

Limitations. Editors' assumption: teachers account for about 70% of Title 2 (staff spending, €59.23bn), hence about €41bn excluding CAS; brought back to pay for active service (41 / 1.15 ≈ €35.7bn), €0.357bn per percentage point (low and central value). The high value adds employer contributions and the teachers' share of the CAS Pensions (70% of €25.16bn), i.e. about €0.59bn per point. “Net” convention (R9), the same as for the public sector pay index point and public sector headcount: at the centre, cost excluding all employer contributions paid to other public administrations (CAS Pensions, employer social contributions); as upper bound, full employer cost. The published “excluding CAS” cost still includes employer contributions other than the CAS: it is brought back to pay for active service alone by the ratio “average loaded cost excluding CAS / pay for active service” from the PAP 2024 (Enseignement scolaire, entry-exit costs: 47,634 / 41,610 = 1.145 in primary, 51,198 / 44,265 = 1.157 in secondary, about 1.15 on average; editors' calculation). At the level of consolidated general government, the CAS is an internal flow: the real additional pension cost comes later. An increase targeted at early careers costs less per average point announced: the 2023 base increase, about 10% « en moyenne » [“on average”], cost €1.9bn. The base targeted by each candidate therefore matters. The pay index point also raises teachers' pay: declared overlap with the “civil service” lever. Parameter in net euros (hausse-nette-mensuelle): 852,752 teachers (712,806 in the public sector, 139,946 in private schools under contract, Senate, PLF 2026) × €1,200 net per year. Conversion from net to gross using the ratio noted by the DEPP for a teacher's bonus (€3,180 gross per year for €227 net per month, i.e. 85.7%): about €1,401 gross per year. Central value, without employer contributions (“net” convention): 852,752 × €1,401 ≈ €1.19bn. Bounds: €100 net is 3.24% of the average net salary of a tenured teacher (€3,090 in 2024, DEPP), i.e. €1.16bn with the low value of the percentage parameter (0.357) and €1.91bn with its high value (0.59), full employer cost with the CAS Pensions share (case of an increase paid on the index-based salary, which accrues pension rights). Limits: the net-to-gross ratio is not the same for the salary as for a bonus; teachers in private schools under contract have different contributions; contract teachers earn less on average (€2,130 net according to the DEPP), which does not change the cost of a uniform increase in euros but makes its conversion into a percentage approximate.

Ability groups, organisation of lower secondary schoolcostable, spending
  • Abolition of needs-based groups, posts returned to the budget: a gain of €0.12bn per year (0.12 to 0.2) (coefficient calculated by the editors from Sénat, rapport spécial PLF 2026, Enseignement scolaire).
  • Extension of ability groups in French and mathematics to 4e and 3e, with the same number of posts as in 6e and 5e (2,800), minus the 340 already assigned to these levels; zero with constant resources (lower bound): a cost of €0.11bn per year (0 to 0.18) (coefficient calculated by the editors from Sénat, rapport spécial PLF 2026, Enseignement scolaire).
Current law, horizon, limitations and 3 sources

Current law. « Choc des savoirs » reform: ability groups in French and mathematics in the first two years of lower secondary school (6e and 5e) since the start of the 2024 school year. Their extension to 4e and 3e was made optional by the minister E. Borne. The scheme drew on 2,800 dedicated posts in 2024-2025.

Horizon. full year 2032, €bn/year

Limitations. Calculation: 2,800 FTE × €48,336 (cost excluding CAS for primary, used for lack of a checked secondary cost; secondary costs more, hence the high value). Extension to 4e and 3e (the last two years of lower secondary school) is a simple extrapolation (same number of classes), without official costing. An abolition “gains” only if the posts are returned to the budget, and not redeployed to reducing class sizes. Secondary cost since checked: €51,198 loaded excluding CAS on average (PAP 2024), i.e. 2,800 FTE × €51,198 ≈ €0.14bn. “Net” convention (R9): at the centre and as lower bound, this cost brought back to pay for active service (51,198 / 1.157 ≈ €44,250, i.e. €0.12bn; 48,336 / 1.145 also gives 0.12); as upper bound, full employer cost with the CAS Pensions (editors' calculation). No official costing of extending the groups to 4e and 3e was found (in 2024-2025, 340 FTE were used there for support, not for groups). Option extension-4e-3e (rule on public bases, 18 October 2026): (2,800 − 340) = 2,460 FTE × €44,250 (secondary cost brought back to pay for active service) ≈ €0.11bn at the centre; 2,460 × €71,400 (full employer cost, CAS Pensions included, that of the upper bound for abolition: €0.2bn / 2,800) ≈ €0.18bn as upper bound; 0 as lower bound if the extension is carried out with constant resources, within the schools' hours (editors' calculations).

Funding of private schoolscostable, spending
  • End of public funding (State and local authorities), static costing without pupils returning to the public sector (theoretical gain), 2022 amount: a gain of €10.4bn per year.
  • Allocation of resources according to social composition, within a constant envelope: a gain of €0bn per year.
  • Share of public funding reduced to 50% of the resources of private schools under contract, with tuition fees in compensation (2022 resources held constant: €6.9bn public instead of 10.4), static costing without pupils returning to the public sector: a gain of €3.5bn per year (coefficient calculated by the editors from Assemblée nationale, rapport d'information n° 2423).
Current law, horizon, limitations and 3 sources

Current law. Loi Debré of 1959: association contracts. The State pays the 142,000 teachers in private schools under contract (programme 139). Local authorities pay the forfait d'externat (per-pupil operating grant), set on the cost of a state-school pupil. More than 2 million pupils, i.e. 17.6% of enrolment.

Horizon. full year 2032, €bn/year

Limitations. The static gain of the first option is fictitious. If pupils return to the public sector, their cost shifts there (private schools cost the State less per pupil), and the net gain may become zero or negative. Amount: €10.4bn of public money in 2022 (€8.5bn from the State and €1.9bn from local authorities, DEPP assessment taken up by information report no. 2423 of the National Assembly); a single source for this scope, no range. Other benchmarks, with other scopes or without a written source, not counted: €8bn for the State alone in 2022 (programme 139, Cour des comptes), more than €9bn for this programme in 2024; « entre 10 et 12 milliards » [“between 10 and 12 billion”] is an oral statement by one of the rapporteurs at a press conference, absent from the report, which judges the real public effort to be « bien supérieur » [“well above”] €10.4bn. Social modulation is cost-neutral if done within a constant envelope. The amount of programme 139 in the PLF 2026 has not been checked (only its 0.7% decrease has). Option financement-public-50-pourcent (rule on public bases, 18 October 2026): in 2022, private schools under contract received €13.8bn, of which €10.4bn of public money, i.e. 75%; bringing this share down to 50% with constant resources, families paying the difference, leaves €6.9bn of public money, i.e. €3.5bn less (editors' calculation, 2022 amounts not updated). Another reading, half of current public funding (€5.2bn), would concern a different object from the share in resources. Same reservation as for abolition: pupils who left for the public sector would shift their cost there.

School timetablesnot costable
Current law, horizon, limitations

Current law. 4.5-day week by default (2013 reform), with an exemption allowing a 4-day week since 2017. A majority of municipalities have returned to 4 days (text and share not cited in the source).

Horizon. Not applicable.

Limitations. The cost falls mainly on municipalities (out-of-school activities) and depends on the existence of a State support fund. No figure read: the lever cannot be costed as long as the cost of the support fund for out-of-school activities is not sourced.

Higher education and research

Access to higher educationnot costable
Current law, horizon, limitations and 1 source

Current law. Loi ORE of 2018: Parcoursup. Non-selective bachelor's programmes (licences) may answer « oui si » [“yes, if”]; master's programmes have been selective since 2017.

Horizon. Not applicable.

Limitations. Abolishing or tightening selection costs nothing in itself. However, a promise of additional places has a cost, which depends on the cost per student (not sourced here); it is costed only if a candidate announces a number of places.

University tuition feescostable, revenue
  • Entirely free registration fees: a cost of €0.5bn per year (0.45 to 0.55).
  • Increase to about €900 for a bachelor's degree and €1,300 for a master's (revenue, Assises du financement, June 2026): a gain of €1.5bn per year (1.2 to 1.5).
  • Abolition of the student and campus life contribution (CVEC): a cost of €0.17bn per year.
  • End of the exemptions from differentiated fees for non-EU students still allowed by the decree of 19 May 2026 (20% of the students concerned in 2028): a gain of €0.07bn per year (coefficient calculated by the editors from IGF-IGESR, Modèle économique des établissements publics de l'enseignement supérieur (vie-publique)).
Current law, horizon, limitations and 6 sources

Current law. Order of 19 April 2019: national fees of €178 for a bachelor's (licence), €254 for a master's and €397 for a doctorate in 2025-2026. Grant holders are exempt. Fees bring in about €500m a year, i.e. 2.7% of institutions' resources. Non-EU students: differentiated fees (€2,895 for a bachelor's, €3,941 for a master's); at the start of the 2024 academic year, of 111,400 students within their scope, 100,400 were fully or partly exempt and 11,100 paid the full rate (SIES); Decree No. 2026-385 of 19 May 2026 lowers the share that can be exempted to 30% at the start of the 2026 academic year, 25% in 2027 and 20% in 2028 (ministerial answer, AN (National Assembly), August 2026).

Horizon. full year 2032, €bn/year

Limitations. The €500m includes non-national fees (university-specific diplomas, non-EU students), so making national degrees alone free costs less. Benchmark: the Institut Montaigne (2022), based on the Cour des comptes, costs free national degrees at universities (excluding engineering schools, 2018-2019 fees) at €342.2m of theoretical fees, minus €112.5m already compensated by the State for grant holders, i.e. about €0.23bn. This benchmark is not counted: the Cour des comptes report could not be opened, and nothing in the sources read indicates that the €500m of institutions' revenue includes the compensation for grant holders. An increase brings in less if the number of exempt grant holders rises, or if the State simultaneously cuts the public service operating grant. In that case, the gain goes to the State and not to the universities: the position specifies this. Option fin-exonerations-droits-differencies (editors' calculation, 18 October 2026): without any exemption, differentiated fees would bring in €334m in 2022-2023 (theoretical yield, Cour des comptes, March 2025); the decree of 19 May 2026 allows at most 20% to be exempted from 2028: ending all exemptions therefore brings in at most 20% × €334m, about €0.07bn per year, as university revenue. Static: the fall in enrolments that higher fees would cause is not counted. The text of the decree could not be read (Légifrance inaccessible): the 20% ceiling comes from a published ministerial answer and from the specialist press. Option suppression-cvec: €170m paid in 2023-2024 by about 1.7 million non-grant-holding students (Cour des comptes). The ceiling set in the finance act was €177m in 2023 and 2024. The cost is the same whether or not the State compensates the Crous and the institutions: the public revenue disappears. The CVEC (student and campus life contribution) is paid across all higher education, not just at university.

Student supportcostable, spending
  • Second phase of the student grant reform (indexation, students living away from home, linearisation): a cost of €0.375bn per year (0.35 to 0.4).
  • +10% on direct support (grants): a cost of €0.25bn per year (0.24 to 0.26) (coefficient calculated by the editors from Sénat, avis PLF 2026 Recherche et enseignement supérieur).
  • Universal autonomy allowance (gross, about €1,000/month): a cost of €27.5bn per year (25 to 30).
Current law, horizon, limitations and 4 sources

Current law. Means-tested grants (bourses sur critères sociaux), not indexed. The highest tier is €6,335/year (unchanged since 2023). 662,000 grant holders in 2024-2025. Programme 231 Vie étudiante (student life): €3.2bn, 78% of it direct aid (PLF 2026). The second phase of the grants reform is suspended. Since 4 May 2026, the €1 meal has been open to all students in all university restaurants, without means testing (€50m from the State in 2026).

Horizon. full year 2032, €bn/year

Limitations. The autonomy allowance would replace part of existing aid (€5.9bn to €6bn: grants, student APL, the additional half-share for tax purposes). Its net cost is therefore about €20bn to €24bn, with an estimation gap between the rapporteur (€30bn) and the minister (€25bn). The +10% option is calculated on 78% of €3.2bn. The €1 meal for all students has been in force since May 2026: a position calling for it is status quo. Its cost (€50m for eight months of 2026 according to the ministry) is part of current policies.

Research, share of GDPcostable, spending
  • Additional public research spending. For +0.1 point of GDP of public research spending in 2032: a cost of €3.5bn per year (2.8 to 3.7) (coefficient calculated by the editors from SIES, notes sur la DIRD 2023 (Flash juillet 2025, NI novembre 2025)). Current value: 0 point of GDP. Range measured by the source: -0.74 to 0.74 point of GDP; beyond it, linear extrapolation is flagged. Arithmetic coefficient (0.1 point of GDP): range from abolition to doubling of government research effort (0.74% of GDP in 2023). Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 3 sources

Current law. The target has been 3% of GDP since 2000. In 2023, gross domestic expenditure on R&D stands at 2.18% of GDP: 1.44% for businesses, 0.74% for government, the lowest level since 1980 (SIES). The research programming act 2021-2030 applies.

Horizon. full year 2032, €bn/year

Limitations. 2023 GDP derived: 61.5 / 0.0218 ≈ €2,820bn, hence 0.1 point ≈ €2.8bn (low value). Nominal GDP in 2032 is projected at €3,500bn-€3,700bn (assumption of about 3% nominal growth per year, not sourced), hence €3.5bn to €3.7bn. If a candidate's objective concerns total GERD (DIRD, 3%), only the State-financed share counts (budget appropriations or research tax credit): the public share is separated from the share expected from businesses. Rule on public bases (18 October 2026), the same for all candidates: a target of 3% of GDP that does not state its public share is read according to the Barcelona objective (European Council, March 2002), under which « les deux tiers de ce nouvel investissement » [“two thirds of this new investment”] come from the private sector: one third of the increase, from 2.18% to 3% of GDP, is public, i.e. (3 − 2.18) / 3 ≈ +0.27 point of public research spending (similar reading: 1% public out of 3%, i.e. +0.26 point). When the text gives its public share, that is what counts. The research tax credit, public funding of business research, is not counted here: it falls under tax expenditures.

Health

Where doctors set up practicenot costable
Current law, horizon, limitations

Current law. Doctors are free to set up practice where they wish. Zoning and incentive payments by the ARS (regional health agencies). Selective agreements with health insurance (conventionnement sélectif) apply to other professions (nurses, physiotherapists). The cross-party Garot private member's bill on regulation was adopted at first reading in the National Assembly in 2025 (parliamentary shuttle not checked).

Horizon. Not applicable.

Limitations. Regulating where doctors set up practice has almost no budgetary cost. Aid for setting up practice and the funding of the 4th year of general practice training can be costed, but are not sourced here.

Public hospitalscostable, spending
  • Increase in health establishments' resources beyond the trajectory. For +1% in resources for the health establishments sub-target (beyond the trajectory): a cost of €1.3bn per year (1.1 to 1.4) (coefficient calculated by the editors from PLFSS 2026, annexe 5 (ONDAM)). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 3 sources

Current law. ONDAM 2026: the « établissements de santé » (health establishments) sub-target funds public and private establishments, at €111.8bn (PLFSS 2026 base). The PLFSS planned +2.4%; in November 2025 the Government announced an €850m increase bringing growth to +3.2% (explanatory statement of amendment no. 2695, which fell at first reading; the final text of the LFSS 2026, which sets the sub-target as voted, has not been reviewed).

Horizon. full year 2032, €bn/year

Limitations. 1% of €111.8bn = €1.12bn today. With about 3% growth per year until 2032 (assumption), the base approaches €137bn, hence €1.3bn to €1.4bn. Commitments of “+X care staff posts” require a cost per FTE in the hospital civil service, not sourced here. The sub-objective includes the for-profit private sector: a position that targets public hospitals specifies this.

Medical deductiblescostable, revenue
  • Total abolition of medical deductibles and flat-rate contributions: a cost of €2.1bn per year (2.1 to 2.3).
  • Return to pre-2024 amounts: a cost of €0.8bn per year (0.7 to 0.8).
Current law, horizon, limitations and 5 sources

Current law. Charges (franchises) of €1 per box of medicine or paramedical procedure and €4 per transport; flat contribution of €2 per consultation or procedure (doubled in March and May 2024). Annual caps raised from €50 to €70 each, i.e. €140 combined instead of €100 (decree no. 2026-858 of 11 September 2026, from 1 October 2026). The €200 plan was abandoned.

Horizon. full year 2032, €bn/year

Limitations. Abolition: central value and lower bound, €1.3bn (stable yield from 2009 to 2020, general scheme, REPSS) + €0.8bn (2024 doubling, government estimate) ≈ €2.1bn; upper bound, €1.5bn (2021 yield, REPSS) + 0.8 ≈ €2.3bn (editors' calculations). The values of €2.4bn and €2.8bn used until 18 October 2026 were not based on any cited source: withdrawn. The 2026 raising of the ceilings (yield not published) is not counted: it would slightly increase the cost of an abolition. A media outlet (Le Tribunal du Net) puts forward €2.5bn in 2025 without a checked primary source. The revenue figures are amounts due: €1.5bn of receivables remained to be collected at the end of 2024 (Cour des comptes, via iFRAP). Notable discrepancy: the PLFSS 2026 costed the initial plan (doubling of amounts and ceilings) at €2.3bn, while Caducée speaks of about €750m for the move to €200 alone.

Extra-billing by doctorscostable, spending
  • Ban or cap without public coverage (budgetary cost): a cost of €0bn per year.
  • Abolition with the amounts covered by national health insurance (Assurance maladie) (cost in 2032): a cost of €5.7bn per year (4.3 to 6.6) (coefficient calculated by the editors from HCAAM, « Les dépassements d’honoraires des médecins : état des lieux », octobre 2025 (pages 4 et 22 du PDF)).
Current law, horizon, limitations and 1 source

Current law. Sector 2 with unregulated fees « avec tact et mesure » [“with tact and moderation”]. OPTAM moderation scheme. 56% of specialists practise in sector 2. Extra fees reach €4.3bn for specialists and €4.5bn for all doctors in 2024, rising 5% a year in real terms over 2019-2024 (HCAAM, 2 October 2025).

Horizon. full year 2032, €bn/year

Limitations. €4.5bn (all doctors, 2024) and +5% per year in real terms over 2019-2024 (+8% in current euros): HCAAM report of October 2025. Central value for 2032: €4.5bn × 1.03^8 (nominal growth of 3%, assumption). High value: €4.5bn × 1.05^8; it applies the real rate of 5% as a nominal rate, and the nominal trend of 8% would give more (calculation: €8.3bn); the range rests on growth assumptions, not on a published costing for 2032. A ban without compensation costs the budget nothing, but may lead doctors to leave the national agreement (déconventionnement) or reduce the supply of care (cannot be costed). Part of the extra fees is already reimbursed by supplementary insurers, and thus not by public spending.

End of lifenot costable
Current law, horizon, limitations and 4 sources

Current law. Act no. 2026-794 of 18 August 2026 on the right to assistance in dying (finally adopted by the National Assembly on 15 July 2026, after three rejections by the Senate; declared constitutional with reservations by the Constitutional Council, decision no. 2026-910 DC of 14 August 2026; JO of 19 August 2026). It allows an adult, French or residing stably and lawfully in France, suffering from a serious and incurable condition that threatens life, at an advanced or terminal stage, with refractory or unbearable suffering, and able to express a free and informed wish, to use a lethal substance that they self-administer (or, if they cannot, administered by a doctor or a nurse). Its implementation depends on implementing decrees (procedure, review committee), which are awaited. Act no. 2026-404 of 26 May 2026 aiming to guarantee equal access for all to palliative care and support. The Claeys-Leonetti act of 2016 (continuous deep sedation until death, advance directives) remains in force.

Horizon. Not applicable.

Limitations. Ethical question, with no direct static budgetary cost: cannot be costed. Only the resources for palliative care (ten-year strategy, law of 26 May 2026) could be, but no official amount has been checked here. Direction of positions: relative to the law of 18 August 2026, supporting the law as voted is status quo, repealing or restricting it a decrease, extending it (minors, advance directives, psychological suffering) an increase; a position predating the vote that calls for « légaliser » [“legalising”] assisted dying remains dated to its day. Implementing decrees: the press says they are expected (Caducée, 25 August 2026); a search on 13 October 2026 found none published, without exhaustive consultation of the Journal officiel.

Housing

Rent controlsnot costable
Current law, horizon, limitations

Current law. Trial under the ELAN act (article 140), extended by the 3DS act, in volunteer municipalities in high-pressure housing areas (Paris, Lille, Lyon, etc.).

Horizon. Not applicable.

Limitations. No direct budgetary effect. An indirect effect on housing benefit (APL) exists (lower rents, hence lower APL) but is not costed here.

Social housingcostable, spending
  • Social housing units financed per year. For 10,000 more social housing units a year: a cost of €0.16bn per year (0.14 to 0.17) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Logement et hébergement d'urgence (F. Jolivet)). Current value: 99,000 homes per year. Range measured by the source: 99,000 to 200,000 homes per year; beyond it, linear extrapolation is flagged. Range measured by the Institut Montaigne: pace raised to 200,000 dwellings a year. Beyond that, or for a decrease, linear extrapolation, flagged as such.
  • Abolition of the RLS (rent reduction in social housing) (cost to the State, which offsets it through APL): a cost of €1.1bn per year (0.9 to 1.3).
Current law, horizon, limitations and 3 sources

Current law. SRU act: 20 to 25% social housing. Solidarity rent reduction (RLS) levied on social landlords since 2018: €1.1bn in 2025, €1.3bn in the PLF 2026, with a compromise target in the Senate of €0.9bn. 99,000 social housing units approved in 2024, about 33,000 of them PLAI (very-low-income housing).

Horizon. full year 2032, €bn/year

Limitations. Coefficient: the Institut Montaigne costs 135,000 additional social housing units per year at €1.9bn to €2.3bn (2.1 central) for general government, i.e. €0.14bn to €0.17bn per 10,000 dwellings (public subsidies of 8 to 9% of a construction cost of €156,000 to €172,000, increased for the RE2020 standard; editors' calculation). “Net” convention: these subsidies include those of Action logement, which is not a public administration; the source does not give its share and the page could not be re-checked in October 2026 (access refused): the coefficient is not net of it and therefore overstates the cost for general government. The reference pace is that of 2024 approvals (99,000); Montaigne started from 65,000 dwellings built. Not counted: the property tax exemption for 25 to 30 years, VAT at 5.5%, the interest rate advantage of loans from the fonds d'épargne (savings fund), land, the equity the State might have to provide (of the order of €2bn according to Montaigne) and the effect on VAT (from −€350m to +€350m depending on whether the dwellings replace or add to other construction). The 5 to 10% increase for the RE2020 standard, specific to the text costed in 2022 (Mélenchon), is applied as is to the other positions. FNAP bricks-and-mortar aid alone (about €5.5k per dwelling) would greatly underestimate the public cost. The RLS (rent reduction scheme) amounts differ according to the stage of the 2026 budget procedure (€1.3bn / €1.1bn / €0.9bn).

Personal housing benefits (APL)costable, spending
  • Uprating of the housing benefit (APL) scale beyond indexation. For +1% uprating of the APL scale (beyond the reference indexation): a cost of €0.16bn per year (0.15 to 0.17) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Logement et hébergement d'urgence). Current value: 0%. Range measured by the source: -100% to 100%; beyond it, linear extrapolation is flagged. Coefficient proportional to the total measured by the source (editors' calculation): arithmetic range from abolition (−100%) to doubling (+100%) of this total. Beyond that, linear extrapolation, flagged as such.
  • Reinstatement of APL for home buyers (APL accession) in metropolitan France: a cost of €0.05bn per year.
  • Reinstatement of APL for non-EU students not receiving grants (full year): a cost of €0.2bn per year.
  • Personal housing benefits reserved for households of French nationality: a gain of €2.71bn per year (1.5 to 2.71) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Logement et hébergement d'urgence).
Current law, horizon, limitations and 7 sources

Current law. Programme 109: €16.1bn in the PLF 2026 (€16.7bn in 2025). Scale frozen in 2026 (« année blanche »), for a saving of €108m. APL abolished for non-EU students who are not grant holders as of 1 July 2026, for a saving of €200m in a full year.

Horizon. full year 2032, €bn/year

Limitations. Option reserve-francais (decision of 18 October 2026, same rule for all candidates who reserve a benefit for French nationals), on personal housing aid spending in 2025 (€16.7bn, PLF 2027) minus the APL (housing benefit) for students from outside the EU who do not hold means-tested grants, already abolished (€0.2bn), i.e. €16.5bn: lower bound, share of foreigners in the population (9.1% in 2025, Insee), €1.50bn; upper bound and central value, share of spending paid to non-EU households (16.4%, 2023 ministerial answer to written question no. 5530, with no reference date, end of 2022 according to other sources), €2.71bn. Nationals of the European Union are not included in this share: for a measure targeting all foreigners, the upper bound remains a minimum. Legal compliance not assessed (principle of equality, European Union law), as elsewhere in the cost schedule. Static calculation: 1% of €16.1bn. Restoring APL for students from outside the EU who do not hold means-tested grants costs €0.2bn in a full year (Senate): dedicated option. Cost schedule coefficient calculated by the editors. The saving from the freeze (€108m) relates to an October uprating, hence to one quarter in 2026: it is not comparable with a full-year cost. APL for home ownership (APL accession): €50m according to the Senate (2019), which does not specify the year of the costing; aid paid over the life of the loan accumulates from one generation of buyers to the next, so the steady-state cost is probably higher. Not updated.

Home ownershipnot costable
Current law, horizon, limitations and 1 source

Current law. Zero-interest loan (prêt à taux zéro) extended to the whole country for new housing (LFI 2025, temporary). Exemption from gift duties for the purchase of a main residence (LFI 2025).

Horizon. full year 2032, €bn/year

Limitations. Lever that can be costed, but the cost of the « prêt à taux zéro » (zero-interest loan) tax credit (Voies et moyens, volume II, PLF 2026) is not recorded here: no coefficient or option (cost of the PTZ, of an additional generation of loans or of an extension to existing homes).

  • AN, rapport spécial PLF 2026, Logement, 2025-10: « Le crédit d'impôt « Prêt à taux zéro », en revanche, va voir son incidence budgétaire croître, en lien avec son extension à tout le territoire pour les logements neufs et le logement collectif en LFI 2025. »
Emergency sheltercostable, spending
  • Additional shelter places. For 10,000 additional shelter places: a cost of €0.1bn per year (0.066 to 0.15) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Logement et hébergement d'urgence). Current value: 0 places. Range measured by the source: -203,000 to 203,000 places; beyond it, linear extrapolation is flagged. Average cost of a place in the existing stock (203,000 places, editors' calculation): arithmetic range from abolition to doubling of the stock. Beyond that, linear extrapolation, flagged as such.
  • Additional annual envelope for emergency accommodation and action against homelessness. For +€1bn per year of appropriations for emergency accommodation or action against homelessness: a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
Current law, horizon, limitations and 2 sources

Current law. Unconditional reception (article L. 345-2-2 of the Code de l'action sociale et des familles). General-purpose capacity stable at 203,000 places since 2021. Programme 177: €3.07bn in payment appropriations in the PLF 2026, of which €1.34bn for general-purpose capacity (emergency shelters, hotels).

Horizon. full year 2032, €bn/year

Limitations. Envelope announced in billions (plan with no number of places): the candidate's amount, taken as is (rule on amounts taken over, decision of 18 October 2026); benchmark: the whole of programme 177 is worth €3.07bn in payment appropriations in the PLF 2026. An envelope and places costed separately would overlap (overlap declared). Low value: €1.34bn / 203,000 ≈ €6.6k per place per year (general-purpose stock). High value: the whole of programme 177 / 203,000 ≈ €15k. The actual cost depends on the type of place: a CHRS (shelter and social reintegration centre) with support costs more than a night in a hotel. The programme is structurally under-budgeted, which makes the “current policies” baseline fragile. People in an irregular administrative situation in the general-purpose stock (excluding the reception scheme for asylum seekers) in 2024: at least 39,000 (low estimate; median 52,000, high 78,000), according to the spending review by the IGF, IGAS and IGA (May 2025), which considers the data fragile (status known for 52% of people).

  • AN, rapport spécial PLF 2026, Logement et hébergement d'urgence, 2025-10: « 1 315,5 millions d'euros en AE et 1 340,3 millions d'euros en CP ... sont consacrés au parc d'hébergement généraliste ... maintien des capacités d'accueil à hauteur de 203 000 places ; 3,047 milliards d'euros en AE et 3,071 milliards d'euros en CP sont prévus au titre du programme 177 »
  • IGF, IGAS et IGA, revue de dépenses sur l'hébergement d'urgence (mai 2025), tableau 13, 2025-05: « Dans le SI-SIAO, le statut administratif des personnes hébergées n'est connu que pour 52 % d'entre elles (soit pour environ 100 000 personnes sur un effectif total de 192 570 places en 2024). […] la Dihal dénombre 39 000 personnes en situation administrative irrégulière. ; tableau 13 : estimation basse 39 000, médiane 52 000, haute 78 000 »

Security and justice

Police and gendarmerie numberscostable, spending
  • Additional police officers and gendarmes. For 1,000 additional police officers or gendarmes (FTEs): a cost of €0.045bn per year (0.043 to 0.075) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Sécurités). Current value: 0 FTE. Range measured by the source: -256,365 to 256,365 FTE; beyond it, linear extrapolation is flagged. Average cost of a police or gendarmerie post (256,365 ETPT in the PLF 2026, editors' calculation): arithmetic range from abolition to doubling of staff numbers. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 1 source

Current law. LOPMI 2023-2027. Police nationale: 153,286 FTEs worked (ETPT); gendarmerie: 103,079 ETPT (PLF 2026). 2026 staffing plan: +1,000 FTEs in the police and +400 in the gendarmerie. Title 2 excluding the pensions special account (CAS): €7.95bn for the police, €5.11bn for the gendarmerie.

Horizon. full year 2032, €bn/year

Limitations. Average cost excluding CAS (special allocation account for pensions): €51.9k per ETPT (full-time equivalent worked) for the police (€7,948.7m / 153,286 ETPT), €49.6k for the gendarmerie, all categories combined, including administrative staff; brought down to activity pay by the ratio of about 1.15 from the 2024 PAP (measured on teachers, applied for lack of a published ratio for the police and gendarmerie; editors' calculation): €45.1k (central value) and €43.1k (lower bound). “Net” convention (R9), the same as for the public sector pay index point and public sector headcount: at the centre, cost excluding all employer contributions paid to other general government bodies (CAS Pensions, employer social contributions); at the upper bound, full employer cost. The published “excluding CAS” cost still includes employer contributions other than the CAS: it is brought down to activity pay alone by the “average loaded cost excluding CAS / activity pay” ratio from the 2024 PAP (Enseignement scolaire, entry-exit costs: 47,634 / 41,610 = 1.145 in primary, 51,198 / 44,265 = 1.157 in secondary, about 1.15 on average; editors' calculation). The former central value (0.06) added 15% for equipment and running costs without a source: removed on 13 October 2026. High value: with the contribution to the CAS Pensions. 2022 benchmarks not counted, with a different scope: iFRAP costed 12,000 police officers at €0.57bn (€47.5k per officer, scope not published); the Institut Montaigne costed a community police force made up three-quarters of contract security assistants at €46k per officer, pensions included. Initial training (one year) delays the effect on the ground.

  • AN, rapport spécial PLF 2026, Sécurités, 2025-10: « Programme 176 Police nationale 147 076 153 286 ; Programme 152 Gendarmerie nationale 99 790 103 079 ; 7 948,7 millions d'euros de crédits de titre 2 hors contribution d'équilibre au CAS Pensions ; 5 111,4 millions d'euros de crédits de titre 2 hors CAS Pensions »
Prisons and sentencescostable, spending
  • Net prison places delivered by 2032. For 1,000 net places delivered by 2032 (annual operation + investment spread over 2027-2032): a cost of €0.11bn per year (0.08 to 0.15) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Justice). Current value: 0 places. Range measured by the source: 0 to 15,000 places; beyond it, linear extrapolation is flagged. Investment cost measured on the « 15 000 places » plan. Beyond 15,000 places, or for a decrease, linear extrapolation, flagged as such.
Current law, horizon, limitations and 4 sources

Current law. At 1 September 2025: 84,311 prisoners for 62,614 operational places, i.e. an occupancy rate of 134.7%. « 15 000 places » plan by 2027, well behind schedule. Modular prisons (target of 3,000 places). Minimum sentences were abolished in 2014.

Horizon. full year 2032, €bn/year

Limitations. Investment: €5.55bn / 15,000 ≈ €370k per place, i.e. about €0.06bn per year for 1,000 places if the works are spread over six years. Running costs: about €40k to €60k per place per year, estimated from Title 2 and the operating appropriations of programme 107. This ratio is overestimated because the programme also funds probation (SPIP). Minimum sentences cannot be costed directly: their cost runs through the number of additional prisoners, hence of places. At 134% prison density, additional prisoners without new places cost the budget almost nothing, but worsen overcrowding. Running costs checked against a source: the Cour des comptes (Court of Audit) (2023) gives €105 per day of detention, i.e. about €38m per year for 1,000 prisoners (running costs only, without investment). With investment spread out (about €0.06bn per year for 1,000 places), this gives about €0.10bn for 1,000 places, within the range used. The Council of Europe (SPACE I 2023) gives €127 per day, cited in a National Assembly document. 15,000 plan: 5,411 net places delivered by autumn 2025; the places still to come are published only in gross terms (about 6,500), which prevents costing a moratorium on the remaining places. Rule on public bases (18 October 2026): a moratorium on new places, excluding replacement places, is costed through the parameter as a negative value (extrapolation, flagged as such) on the gross places not yet under construction (2,560 at design-study stage, 1,550 with contracts awarded, i.e. 4,110), converted into net places at the ratio observed for the facilities delivered (5,411 net for 7,384 gross, 73%, which takes out the closures of replaced prisons): about 3,000 net places (editors' calculation; the 2,388 gross places under construction are treated as committed).

Juvenile justicecostable, spending
  • Additional places in closed educational centres. For 100 additional places in closed educational centres (operation): a cost of €0.021bn per year (0.017 to 0.025) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Justice). Current value: 0 places. Range measured by the source: from 0 places; beyond it, linear extrapolation is flagged. Average daily cost of a place in 2026; the source does not give the number of existing places: no measured upper bound.
Current law, horizon, limitations and 1 source

Current law. Code de la justice pénale des mineurs (juvenile criminal justice code, 2021). Closed educational centres: cost of €578 per place per day in 2026 (public sector).

Horizon. full year 2032, €bn/year

Limitations. Calculation: 100 × €578 × 365 days = €21m. Low value based on a UEHC (residential educational unit), high value with an occupancy rate and investment. Most of the lever (age of criminal majority, immediate trial, mitigation of responsibility) is not budgetary: cannot be costed apart from placement facilities.

  • AN, rapport spécial PLF 2026, Justice, 2025-10: « En 2026, le coût d'une place par jour et par type de structure du secteur public serait le suivant : 578 euros en centre éducatif fermé (CEF), 510 euros en centre éducatif renforcé (CER) et 467 euros dans une unité éducative d'hébergement collectif (UEHC). »
Resources for the justice systemcostable, spending
  • Additional posts in the judicial services. For 1,000 additional posts in the judicial services (judges, court clerks, attachés): a cost of €0.05bn per year (0.05 to 0.09) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Justice). Current value: 0 FTE. Range measured by the source: -39,482 to 39,482 FTE; beyond it, linear extrapolation is flagged. Average cost of a post in programme 166 (39,482 ETPT, editors' calculation): arithmetic range from abolition to doubling of staff numbers. Beyond that, linear extrapolation, flagged as such.
  • Justice mission appropriations excluding the pensions special account (CAS Pensions). For €1bn of additional appropriations for the Justice mission (excluding the contribution to the CAS Pensions): a cost of €1bn per year (candidate’s amount, taken as is). Current value: 10.63 €bn.
Current law, horizon, limitations and 2 sources

Current law. LOPJ 2023-2027: +10,000 FTEs by 2027, including 1,500 judges and 1,800 court clerks. Justice mission: €13.05bn in payment appropriations, €10.63bn excluding the pensions special account (CAS Pensions) (PLF 2026). 9,007 judges in courts in 2025.

Horizon. full year 2032, €bn/year

Limitations. Average cost excluding CAS (special allocation account for pensions) for programme 166: €2.3bn / 39,482 ≈ €58k per ETPT (full-time equivalent worked); brought down to activity pay by the ratio of about 1.15 from the 2024 PAP (measured on teachers, applied for lack of a published ratio for the judicial services; editors' calculation): about €50k, low and central value. A magistrate costs more (high value, full employer cost with the CAS). “Net” convention (R9), the same as for the public sector pay index point and public sector headcount: at the centre, cost excluding all employer contributions paid to other general government bodies (CAS Pensions, employer social contributions); at the upper bound, full employer cost. The published “excluding CAS” cost still includes employer contributions other than the CAS: it is brought down to activity pay alone by the “average loaded cost excluding CAS / activity pay” ratio from the 2024 PAP (Enseignement scolaire, entry-exit costs: 47,634 / 41,610 = 1.145 in primary, 51,198 / 44,265 = 1.157 in secondary, about 1.15 on average; editors' calculation). For a commitment expressed as a budget (“double the justice budget”), take the difference in €bn on the basis of €10.63bn excluding CAS. “Mission budget” parameter: cost equal to the difference between the target amount and €10.63bn (2026 appropriations excluding CAS Pensions, Senate). It covers the whole mission (judicial justice, prison administration, PJJ, access to law): it is not added to a costing of prison places or of closed educational centres for the same candidate. It is used for commitments expressed as a total budget, not for recruitment, which is costed by the jobs parameter.

  • AN, rapport spécial PLF 2026, Justice, 2025-10: « Hors contribution au CAS Pensions, les dépenses de personnel s'établiraient à 2,3 milliards d'euros ... pour un plafond d'emplois fixé à 39 482,35 ETPT ; Nombre total de magistrats exerçant en juridictions ... 9 007 ; LOPJ à 10 000 équivalents temps plein (ETP) d'ici à 2027, dont 1 500 magistrats et 1 800 greffiers »
  • Sénat, rapport général n° 139 (2025-2026), tome III, annexe 17 « Justice » (A. Lefèvre), 2025-11-24: « « En 2026, le projet de loi de finances prévoit que les crédits, hors contribution au compte d'affectation spéciale « Pensions » et à périmètre constant, s'établirait à un niveau de 10 629 millions d'euros, en hausse de 166 millions d'euros en crédits de paiement ou 1,6 % » »
Violence against women and childrencostable, spending
  • Additional public funding against violence against women and children. For +€1bn a year of public funding against violence against women and children: a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
Current law, horizon, limitations and 6 sources

Current law. Interministerial plan for equality between women and men 2023-2027, whose first strand is devoted to violence against women; plan to combat violence against children 2023-2027 (€11.5m in the PLF 2027). Programme 137 « Égalité entre les femmes et les hommes »: €98.65m in the LFI 2026, €106.28m in the PLF 2027. 2026 cross-cutting policy document: €562.32m in payment appropriations for the strand « Combattre les violences faites aux femmes » (combating violence against women). The Government puts existing spending against sexual and gender-based violence at €1.5bn (€1bn from the State, €0.5bn from social security). The « intégrale » (comprehensive) private member's bill against sexual, gender-based and domestic violence was adopted at first reading by the National Assembly on 13 October 2026 (accelerated procedure); the Senate has not yet examined it. Its article 68 schedules €3,000m in appropriations a year from 2027 to 2032.

Horizon. Spending in a full year, static: €1 announced = €1 of deficit, compared with current policies, under which the programming of the comprehensive law has not been finally passed.

Limitations. Envelope parameter, like public investment or rail: it applies only to an amount announced on top of current resources. A commitment expressed as a total budget (“raise the resources to X”) comes down to the difference from existing spending, which the Government puts at €1.5bn for sexual and gender-based violence; this amount aggregates appropriations from several budgets and is not published line by line (the 2026 cross-cutting policy document puts the « violences » (violence) axis at €562.32m; the 2027 one is not yet published). If the comprehensive law is finally adopted with its programming, the current-policies baseline changes and this cost schedule changes: the Government estimates the cost of its measures, eventually, at €3bn to €4bn on top of existing spending. Benchmark, not counted: the Fondation des Femmes (2023, non-profit source) estimates the needs at €2.6bn to €5.4bn per year for violence against women alone. Part of these resources may also appear in the Justice budget (victim support) or in police headcount: overlap declared with the Justice mission budget.

State and civil service

Numbers and pay of civil servantscostable, spending
  • Increase in the value of the public sector pay index point (all three branches). For +1% in the value of the index point (all three branches); central value excluding contributions paid to other public administrations (net convention); upper bound, gross employer cost, about €2.1bn to €2.3bn per point: a cost of €1.53bn per year (1.47 to 2.27) (coefficient calculated by the editors from Sénat, « L'essentiel » sur le PLF 2026, programme « Fonction publique » de la mission « Transformation et fonctions publiques » (page 2)). Current value: 0%. Range measured by the source: 0% to 13%; beyond it, linear extrapolation is flagged. Range measured by the Cour des comptes (note of September 2026, chart no. 11 and table no. 4): from 0 to 100% indexation of the point on inflation of 1.75% a year from 2026 to 2032, i.e. up to about 13% cumulative increase. Beyond that, or for a cut in the point, linear extrapolation, flagged as such.
  • More public employees (or fewer, negative value). For 10,000 public employees: €0.42bn at the average gross salary excluding employer contributions (€244.6bn / 5.88 million employees), €0.62bn at the full employer cost, €0.28bn for local government employees (Cour des comptes): a cost of €0.42bn per year (0.28 to 0.62) (coefficient calculated by the editors from Sénat, « L'essentiel » sur le PLF 2026, programme « Fonction publique » de la mission « Transformation et fonctions publiques » (page 2)). Current value: 0 staff. Range measured by the source: -150,000 to 150,000 staff; beyond it, linear extrapolation is flagged. Range measured by the Cour des comptes: 140,000 fewer local government employees by 2032, rounded to ±150,000. Beyond that, linear extrapolation, flagged as such.
  • Public sector pay index point indexed to inflation each year (editors' common assumption: 1.75% per year, 1.5 to 2.0%; five upratings from 2028 to 2032, i.e. a cumulative increase of 9.1% in 2032, 7.7 to 10.4%), compared with the frozen point; reading key: a one-point gap for one year costs €1.53bn (1.47 to 2.27): a cost of €13.86bn per year (11.36 to 23.63) (coefficient calculated by the editors from Sénat, « L'essentiel » sur le PLF 2026, programme « Fonction publique » de la mission « Transformation et fonctions publiques » (page 2)).
  • 10% increase in the pay of category B and C staff (all three branches of the civil service): category B and C civil servants only at the lower bound, all statuses at the central value, full employer cost at the upper bound: a cost of €12.72bn per year (8.77 to 18.83) (coefficient calculated by the editors from DGAFP, rapport annuel 2026 sur l'état de la fonction publique, et Cour des comptes, note « La masse salariale » (2026)).
Current law, horizon, limitations and 4 sources

Current law. Public sector pay index point frozen at €4.9228 per month since 1 July 2023. 5.88 million public employees at the end of 2024. General government wage bill: €362.1bn in 2024 (21.7% of public expenditure), of which €244.6bn in gross wages and salaries and €117.6bn in employer contributions. The Cour des comptes (September 2026) recommends under-indexing the point and cutting local government staff by 140,000 by 2032.

Horizon. full year 2032, €bn/year

Limitations. Pay index point. Cost of one point in a full year, whole civil service (Senate, opinion on the PLF 2026): 3.5% on 1 July 2022 = €7.6bn, i.e. €2.17bn per point; 1.5% on 1 July 2023 = €3.4bn, i.e. €2.27bn per point (2023 euros, latest observed cost). These costs are read as full employer costs, employer contributions included (editors' reading: the Senate does not detail their composition). “Net” convention of the cost schedule, the same for all public sector staff levers: at the centre, cost excluding employer contributions, a large part of which (CAS Pensions, CNRACL, social security funds) goes back to other general government bodies; at the upper bound, full employer cost. Central value: 2.27 × 244.6 / 362.1 = €1.53bn per point, share of gross wages and salaries (€244.6bn) in the 2024 public sector wage bill (€362.1bn, Cour des comptes); lower bound, same calculation on the 2022 cost: 2.17 × 244.6 / 362.1 = 1.47; upper bound, 2.27, full employer cost for 2023 (editors' calculations). If the Senate figures did not include employer contributions, the central value would underestimate the cost by about a third, a gap covered by the upper bound. According to the Cour des comptes (note of September 2026, chart no. 11 and table no. 4), full indexation of the point on 1.75% inflation per year from 2026 to 2032 would cost €26.6bn in 2032, i.e. €2.1bn to €2.4bn per point of cumulative increase depending on whether seven or six upratings are counted (editors' calculation): a cost of the same order as the Senate's, cited, not counted. The same note costs under-indexation: 25% → €6.7bn, 50% → 13.3, 75% → 20.0, 100% → €26.6bn of savings in 2032. The CGT puts forward a net cost of €1.2bn to €1.4bn after the return through contributions and tax: this costing is not static in the sense of the method. The point also raises teachers' pay, and an increase in the SMIC raises the lowest salaries: overlaps declared. Headcount: “net” convention of the cost schedule, central value at the average gross salary excluding employer contributions, which go back to other general government bodies (€244.6bn of gross wages and salaries for 5.88 million employees, i.e. about €41,600 per employee and €0.42bn for 10,000); upper bound at full employer cost (€362.1bn, about €61,600 per employee, €0.62bn); lower bound, €0.28bn for 10,000 local government employees according to the Cour (€3.9bn / 140,000). Non-replacement of departures: the Cour des comptes (same note, table no. 3) costs the non-replacement of 10% to 50% of departures at €3.6bn to €18.2bn of savings in 2032, covering all public employees, without publishing the number of departures in that scope (only those of the State civil service are published: 408,000 from 2026 to 2032, table no. 1): no cost per unreplaced post can properly be derived from it, and positions that rely on unreplaced departures remain costed at the average cost of an employee (check of 18 October 2026). For a local government scope, the lower bound (€0.28bn for 10,000 employees) is the Cour's coefficient. An overall headcount target may overlap the job creations costed on the sector levers (teachers, police and gendarmerie, justice): overlaps declared, with a zero common base by default, as no source says that an overall headcount target is counted net of these creations. Indexation of the point to inflation (option, rule on public bases of 18 October 2026): common editors' assumption for all indexation rules (pensions, pay index point, grants), inflation of 1.75% per year at the centre (Cour des comptes assumption, note of September 2026), 1.5 to 2.0% for the range; first effect of the term on 1 January 2028, cost in 2032 = cumulative gap of five upratings (2028 to 2032) with the current rule, a frozen point (no uprating since 1 July 2023). Cumulative increase: 1.0175^5 − 1 = 9.06% (1.015^5 − 1 = 7.73%; 1.02^5 − 1 = 10.41%), multiplied by the cost of one point (€1.53bn at the centre, 1.47 at the lower bound, 2.27 at the upper bound): €13.86bn (11.36 to 23.63; editors' calculations, 2023 euros for the cost of the point, as for the parameter). Reading key: one point of indexation gap for one year is worth the cost of one point, €1.53bn. Benchmark, not counted: the Cour des comptes costs full indexation on 1.75% from 2026 to 2032 (six or seven upratings) at €26.6bn in 2032. 10% increase for categories B and C (option): in 2024, the civil service has 39% category A staff, 23% category B and 38% category C, all statuses (DGAFP, 2026 annual report, p. 25); category B civil servants (16% of staff) earn on average €2,657 net per month, category C ones (26%) €2,145, against an average of €2,742 (p. 100). Share of categories B and C in total pay: civil servants only, (16 × 2,657 + 26 × 2,145) / (100 × 2,742) = 35.8% (lower bound); all statuses, attributing to category B and C contract staff the average salary of civil servants in their category, (23 × 2,657 + 38 × 2,145) / (100 × 2,742) = 52.0% (centre; contract staff earn €2,206 on average, slightly less: a slight overestimate). Cost: 10% × share × €244.6bn of gross wages and salaries (net convention, Cour des comptes 2024), i.e. €8.77bn and €12.72bn; upper bound, 10% × 52.0% × €362.1bn of full employer cost, €18.83bn (editors' calculations). Limitations: the headcount shares include military personnel, the salaries do not; the salary scope is in full-time equivalents. Teachers, almost all in category A, are costed under their own lever. An increase in the SMIC also raises the lowest category C salaries: overlap not estimated.

  • Sénat, « L'essentiel » sur le PLF 2026, programme « Fonction publique » de la mission « Transformation et fonctions publiques » (page 2), 2025-11-25: « Face à la crise inflationniste née du conflit en Ukraine, le Gouvernement a mis en œuvre, au 1er juillet 2022, une première revalorisation de 3,5 % du point d’indice[6], dont le coût est évalué à 7,6 milliards d’euros en année pleine pour l’ensemble de la fonction publique, puis une deuxième revalorisation de 1,5 % à compter du 1er juillet 2023, pour un coût de 3,4 milliards d’euros en année pleine. »
  • Cour des comptes, note structurelle « La masse salariale », 21 septembre 2026 (pages 7, 28 et 32 du PDF), 2026-09-21: « la masse salariale brute des administrations publiques, soit 362,1 Md€ ou 21,7 % des dépenses publiques totales en 2024, a progressé de 89,5 Md€ depuis 2012. […] Stabiliser puis rapprocher progressivement d’ici 2032 les effectifs des collectivités territoriales le plus possible de leur niveau de 2014 (soit une diminution de 140 000 agents), notamment au sein du « bloc communal » […] permettrait la réalisation d’une économie maximale de 3,9 Md€. […] Tableau n° 4 : impact budgétaire en 2032 de plusieurs scénarii d’indexation de la valeur du point d’indice / Taux de sous-indexation Économie en 2032 / 25 % 6,7 Md€ / 50 % 13,3 Md€ / 75 % 20,0 Md€ / 100 % 26,6 Md€ »
  • Cour des comptes, note structurelle « La masse salariale », 21 septembre 2026 (pages 7, 17 et 30 du PDF), 2026-09-21: « Au 31 décembre 2024, la fonction publique employait 5,88 millions d’agents […] La masse salariale des administrations publiques (APU) s’est élevée en 2024 à 362,1 Md€ (soit 244,6 Md€ de salaires et traitements bruts et 117,6 Md€ de cotisations employeurs). […] Graphique n° 11 : économies induites selon le taux d’indexation sur l’inflation (avec une valeur prévisionnelle annuelle de 1,75 %) de la valeur du point d’indice entre 2026 et 2032 (trois versants de la fonction publique) »
  • DGAFP, Rapport annuel sur l'état de la fonction publique, édition 2026, « Faits et chiffres » (pages 25 et 100), 2026-10: « « En 2024, la fonction publique est composée de 39% d'agents de catégorie A, de 23% d'agents de catégorie B et de 38% d'agents de catégorie C » ; « Les fonctionnaires de catégorie C constituent 26% de l'ensemble des agents et perçoivent un salaire net moyen mensuel de 2145 euros. Celui des fonctionnaires de catégorie B (16% des agents) vaut 2657 euros. Enfin, le salaire net moyen des fonctionnaires de catégorie A (30% des agents) s'élève à 3460 euros par mois. […] Le salaire net mensuel moyen des contractuels est inférieur à celui des fonctionnaires, et atteint 2206 euros. » ; « Le salaire net moyen dans la fonction publique est de 2742 euros » »
Delegation of public services to the private sectornot costable
Current law, horizon, limitations

Current law. Use of the private sector, for-profit or not, for care homes (EHPAD), crèches, laboratories, etc., under the authorisation and pricing regimes specific to each sector.

Horizon. Not applicable.

Limitations. The cost of a “remunicipalisation” or of bringing services back in-house depends on the sector, on the buyback of assets and on public/private cost differences, which are not documented in general. To be dealt with case by case (care homes (EHPAD), crèches) in the relevant areas.

Public broadcastingcostable, spending
  • Privatisation of France Télévisions and Radio France (annual budgetary gain, excluding sale proceeds): a gain of €3bn per year (3 to 3.09) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Médias, livre et industries culturelles ; Avances à l'audiovisuel public (D. Masséglia)).
  • Abolition of all public funding of public broadcasting: a gain of €3.75bn per year (3.75 to 3.88) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Médias, livre et industries culturelles ; Avances à l'audiovisuel public (D. Masséglia)).
  • Return to an earmarked levy, on a constant scope (neutral if it replaces the VAT fraction): a gain of €0bn per year.
  • Privatisation of « l'audiovisuel public » [“public broadcasting”] with no specified scope: at least France Télévisions and Radio France (€3.0bn net), at most all the companies (€3.88bn gross), annual budgetary gain excluding sale proceeds: a gain of €3bn per year (3 to 3.88) (coefficient calculated by the editors from AN, rapport spécial PLF 2026, Médias, livre et industries culturelles ; Avances à l'audiovisuel public (D. Masséglia)).
Current law, horizon, limitations and 3 sources

Current law. Licence fee abolished in 2022. Funding through a share of VAT allocated to the « Avances à l'audiovisuel public » account (organic law allowing it to continue beyond 2024): €3,878.3m in the PLF 2026 (−1.8%). France Télévisions €2,440.6m, Radio France 648.0, Arte France 298.1, France Médias Monde 303.9, INA 103.5, TV5 Monde 84.2.

Horizon. full year 2032, €bn/year

Limitations. The gain from privatisation is a static costing: it ignores transition costs (redundancy plans, takeover of commitments, broadcasting contracts), any compensated public service obligations and the sale proceeds, which are received only once and do not count in the annual balance. “Net” convention: €126.2m of the 2026 grant offsets tax effects linked to the abolition of the licence fee, which go back to the State (€54.4m for France Télévisions, €30.3m for Radio France); the central value deducts them (3,878.3 − 126.2 ≈ €3,752m for the abolition of all funding; 2,440.6 + 648.0 − 54.4 − 30.3 ≈ €3,004m for France Télévisions and Radio France, editors' calculations), the upper bound is the gross amount. The two options overlap (overlap declared). A privatisation of “public broadcasting” with no list of the companies concerned (same reading for all candidates, rule of 18 October 2026) is costed through the option privatisation-perimetre-non-dit: at the centre and as lower bound, France Télévisions and Radio France, the two main companies (€3.0bn net, 80% of the net grant); as upper bound, all the companies (€3.88bn gross). The final amount in the initial finance act for 2026 (LFI 2026) is not verified (amendments to restore funding under discussion).

Transition and economy

Climate

Emissions cut by 2030, compared with 1990not costable
Current law, horizon, limitations and 3 sources

Current law. Code de l'énergie, art. L100-4, I, 1° (version in force since 24 June 2023, amended by act no. 2023-491 of 22 June 2023): 40% reduction in GHG emissions between 1990 and 2030, carbon neutrality in 2050. The SNBC-3 (decree no. 2026-636 of 16 July 2026, JO of 18 July 2026) sets carbon budgets of 342 Mt/year (2024-2028), 262 Mt/year (2029-2033) and 194 Mt/year (2034-2038), excluding sinks. At EU level, the commitment is -55% net. The Gremillet private member's bill planned to move from 40% to 50% gross; it had not amended the article at the date of reading (13 October 2026, no later version on Légifrance).

Horizon. Not applicable.

Limitations. A target has no cost of its own: the cost runs through the instruments (renovation, rail, support for renewables, public investment). Costing them here as well would be double counting. As a benchmark only: according to I4CE (2024), reaching -55% in 2030 requires €71bn per year of additional public spending by 2030, with regulation and taxation unchanged. The figure falls to about €39bn, of which €18bn for the State, if regulation is tightened and aid refocused. In 2024, public authorities spent €32bn on climate. The I4CE figure is a need, not a cost relative to current policies.

Carbon pricecostable, revenue
  • Carbon component of excise duties. For +€10/tCO2 on the carbon component: a gain of €1.3bn per year (0.9 to 1.9) (coefficient calculated by the editors from La composante carbone en France (I4CE, Point climat n°56)). Current value: 44.6 €/tCO2. Range measured by the source: 30.5 to 86.2 €/tCO2; beyond it, linear extrapolation is flagged. Range measured: 2017-2022 trajectory of the carbon component, from €30.5 to €86.2/t. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 4 sources

Current law. The carbon component of excise duties on energy (formerly TICPE, TICGN, TICC) has been frozen at €44.6/tCO2 since 2018. The trajectory of the LFI 2018 (€55 in 2019, €86.2 in 2022) was abandoned after the gilets jaunes. The European ETS2 market (road fuels, buildings; Directive (EU) 2023/959) is postponed to 1 January 2028, with a price stability mechanism around €45/t (the ceiling is under revision in 2026).

Horizon. Gross revenue in a full year, without behavioural effects, excluding induced VAT and excluding compensation (energy cheque, etc.). Fossil energy volumes decline by 2032 (SNBC-3), so the lower bound is the most prudent for 2032.

Limitations. The cost schedule is derived from I4CE figures and is not a recent official costing. High slope: raising the rate from €30.5/t to €44.6/t in 2018 was expected to bring in €2.7bn, i.e. about €0.19bn per €/t. Low slope: the 2017-2022 trajectory (€5.4bn at €30.5/t, €10.8bn planned at €86.2/t) gives about €0.097bn per €/t. Central value: average yield at €86.2/t, i.e. about €0.125bn per €/t. No official source from 2023 to 2026 isolates the current yield of the carbon component. Connaissance des énergies speaks of « avoisineraient 10 milliards » [“would be close to 10 billion”], without a source. The arrival of ETS2 in 2028 on the same base may lead to a reduction in the national component, which creates a risk of double counting. ETS2 revenue goes partly to the EU Social Climate Fund. The allocation of the revenue is read position by position: in most proposals, it funds redistribution. The return to the 2018 trajectory and abolition are given by the linear parameter.

Energy renovations per yearcostable, spending
  • Major renovations assisted per year. For +100,000 major renovations assisted per year: a cost of €2.6bn per year (1.2 to 3.6) (coefficient calculated by the editors from Chiffres clés de l'Anah 2024 (DREAL Corse)). Current value: 91,374 homes. Range measured by the source: 91,374 homes only; beyond it, linear extrapolation is flagged. A single year measured by Anah (2024: 91,374 major renovations, average aid per dwelling); any other value is a linear extrapolation, flagged as such.
  • +100,000 single-measure renovations per year (average grant €3,622): a cost of €0.36bn per year (coefficient calculated by the editors from Chiffres clés de l'Anah 2024 (DREAL Corse)).
Current law, horizon, limitations and 2 sources

Current law. The main instrument is MaPrimeRénov' (Anah). It comprises an assisted pathway for major renovations, per-measure aid and aid for co-owned buildings. It is supplemented by energy savings certificates (CEE). In 2024: 91,374 major renovations (€2,386m in aid) and 340,801 dwellings assisted in total (€3.29bn). The scheme was suspended from June to September 2025, then reopened on 23 February 2026 after the LFI 2026. The final amount of 2026 appropriations differs between sources (€3.5bn or €4.3bn in Anah commitments) and has not been checked in the act.

Horizon. Direct public aid in a full year. Anah 2024 scale in current euros. Includes neither the CEE (energy saving certificates), funded by energy suppliers, nor the cost of support services.

Limitations. Central value: average aid across all major renovations (€2,386m / 91,374 ≈ €26,100). Lower bound: average aid in jointly owned buildings (€11,818). Upper bound: individual supported pathway (€36,271). In 2024, 59% of households aided for a major renovation were very low-income: a higher target would reach better-off households, with lower unit aid but an uncertain take-up rate. Fraud justified the 2025 suspension. A target announced in “renovated” homes is converted into major renovations before applying this cost schedule. Only the 2024 financial year is published by Anah in the sources read (key figures); 2025 volumes are known only from the press: the parameter's domain is limited to 2024.

End of sales of combustion-engine carsnot costable
Current law, horizon, limitations and 1 source

Current law. The European regulation on CO2 standards for new cars (2023) requires a 100% cut in emissions in 2035, which amounts to banning the sale of new combustion-engine cars. On 16 December 2025, the Commission proposed lowering the target to -90% compared with 2021, with offsetting of the remaining 10%. The proposal was still under negotiation according to the sources read.

Horizon. Not applicable.

Limitations. This is a European standard, with no direct budgetary effect in France. Budgetary costs run through other levers: the bonus (now funded by the CEE, energy savings certificates), social leasing, the scrappage bonus, and eventually the erosion of excise duties on fuels. A national position does not change the law, which is decided by qualified majority in the Council and the European Parliament.

Adaptation to climate changecostable, spending
  • Additional public resources for climate change adaptation. For +€1bn a year of public spending on climate change adaptation: a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
Current law, horizon, limitations and 6 sources

Current law. Third national climate change adaptation plan (PNACC-3), announced on 10 March 2025, which adopts a reference warming trajectory of +4 °C by 2100. Fonds de prévention des risques naturels majeurs (major natural hazards prevention fund, « fonds Barnier », programme 181): €300m in commitment authorisations and €228.8m in payment appropriations in the LFI 2026; €300m and €238.8m in the PLF 2027. Fonds vert (green fund): €1bn in commitment authorisations and €981.4m in payment appropriations in the PLF 2027, including a dedicated adaptation envelope of €500m in commitment authorisations.

Horizon. Spending in a full year, static: €1 announced = €1 of deficit, compared with current policies.

Limitations. Envelope parameter, like public investment or rail: it applies only to an amount announced on top of current resources. A fund announced as a total amount that brings together existing appropriations is counted by the gap with those appropriations, in payment appropriations of the PLF 2027 (Barnier fund €238.8m, Fonds vert €981.4m), the reading in commitment appropriations (€300m and €1bn) being cited in the note. Adaptation spending is scattered (Barnier fund, Fonds vert, civil protection, agriculture, health, local authorities) and has no official total: the Cour des comptes (2024) notes « l'absence de chiffrages exhaustifs et cohérents pour l'ensemble des acteurs publics » [“the absence of exhaustive and consistent costings for all public actors”], and I4CE (2025) identifies €1.7bn explicitly dedicated to adaptation in 2025. Benchmarks, not counted: the National Assembly's flash fact-finding mission on summer 2026 (rapporteur Éric Coquerel) estimates the cost of that summer at about €16bn and puts the public spending needed from summer 2027 at between €10bn and €13.6bn; these are estimates, not coefficients. A plan that funds investment works may overlap a public investment envelope (overlap declared); a plan that strengthens hospitals or care homes (EHPAD) is not also counted on the hospital or old-age levers.

Energy

New nuclear reactorscostable, spending, weighs on the debt
  • EPR2 reactors committed. For 1 EPR2 reactor more (or fewer) than the 6 in the PPE3 — State loan disbursed per year during construction: a cost of €0.73bn per year (0.61 to 1.2) (coefficient calculated by the editors from EDF, communiqué du 18 décembre 2025, « EDF shares its forecasted cost estimate of the EPR2 programme for €72.8bn » (copie d'archive du 13 octobre 2026, le site renvoyant une erreur 403)). Current value: 6 reactors. Range measured by the source: 0 to 6 reactors; beyond it, linear extrapolation is flagged. The coefficient assumes disbursements spread from 2027 to 2038: it holds for the 6 EPR2 reactors of the PPE3. The 8 optional reactors (2045 horizon) would not be in full disbursement in 2032: beyond 6, linear extrapolation, flagged as such, overestimates the cost.
Current law, horizon, limitations and 4 sources

Current law. The PPE3 (decree published in the JO on 13 February 2026, period 2026-2035) provides for 6 EPR2 reactors (Penly, Gravelines, Bugey), with a first commissioning targeted in 2038 and 8 additional reactors as an option. It removes early reactor closures. EDF's estimate of 18 December 2025 is €72.8bn (2020 euros, excluding financing costs). The financing planned by the Nuclear Policy Council in March 2025 rests on a subsidised State loan covering at least half of construction and a 40-year CfD capped at €100/MWh in 2024 euros; the Council of 12 March 2026 set this loan at 60% of the total programme amount, funded by the Caisse des dépôts savings fund, and EDF committed to keeping to €72.8bn (2020 euros). It still has to be approved by the European Commission. The final investment decision is expected at the end of 2026.

Horizon. In €bn per year of State financing requirement (debt), during construction (about 2027-2038). The Maastricht deficit is affected only by the loan subsidy, which is not costed by the sources read.

Limitations. This cost is not of the same nature as the others: a loan is a financial transaction that increases debt but not the deficit, except for the subsidised portion and any call on the CfD. Calculation: lower bound 72.8/6 ≈ €12.1bn per reactor (EDF, 2020 euros) × 60% / 12 years ≈ 0.61; central 79.9/6 ≈ €13.3bn (Cour des comptes, 2023 euros, EDF costing of late 2023, since superseded by that of December 2025) × 60% / 11 years ≈ 0.73; upper €13.3bn entirely lent over 11 years. Share lent: 60% (Conseil de politique nucléaire of 12 March 2026). The durations of 11 to 12 years are the editors' assumptions (from the decision at the end of 2026 to commissioning in 2038); yet EDF is aiming for 2038 for the first reactor and an interval of 12 to 18 months between the following ones: for the last reactors, disbursement in 2032 is probably lower, and the coefficient overestimates it. The loan will be funded by the savings fund of the Caisse des dépôts: its effect on State debt remains to be established. The Cour des comptes (January 2025) noted the absence of a « devis abouti » [“finalised estimate”]. Greenpeace estimates the relaunch at between €380bn and €650bn (campaigning figure, different scope). The CfD may cost or bring in money depending on market prices after 2038, hence beyond the 2032 horizon. Sign convention for options: negative = gain. Abandonment is given by the linear parameter. The 8 optional reactors are no longer an option: their disbursement schedule (2045 horizon) is not sourced, and the coefficient, calculated over 2027-2038, would overestimate them in 2032.

Wind and solarcostable, spending
  • Renewable electricity supported (change). For +1 TWh of renewable electricity supported per year (or -1 TWh for a moratorium): a cost of €0.091bn per year (0.089 to 0.092) (coefficient calculated by the editors from Délibération CRE n°2026-149 du 15 juillet 2026 (CSPE 2026-2027)). Current value: 0 TWh/year. Range measured by the source: -98 to 98 TWh/year; beyond it, linear extrapolation is flagged. Average unit extra cost forecast by the CRE for 98 TWh supported in 2027 (editors' calculation): range from the end of support for this volume to its doubling. Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 5 sources

Current law. PPE3 targets (decree of February 2026): solar photovoltaic 48 GW in 2030 and 55 to 80 GW in 2035; onshore wind 31 GW in 2030 and 35 to 40 GW in 2035; offshore wind 15 GW in 2035. Support is provided through purchase obligations and additional remuneration, offset by the State budget (CSPE). According to the CRE (deliberation 2026-149 of 15 July 2026), the charges to be offset amount to €14.2bn in 2027, of which €10.7bn entered in the State budget, for 98 TWh supported in 2027 and a unit extra cost of €91.2/MWh.

Horizon. Public service charges (State budget) in a full year, at the average unit extra cost observed or forecast by the CRE for 2025-2027.

Limitations. The average additional cost includes the old photovoltaic contracts from before the 2010 moratorium, which cost nearly €2bn per year and mostly expire before 2033. New contracts have a lower marginal additional cost, which the CRE (energy regulator) does not isolate: the cost schedule therefore probably overestimates the cost of a new TWh. The cost depends heavily on the wholesale price: during the 2022-2023 crisis, renewables brought €5.5bn into the budget. A wind moratorium only saves future contracts, not the existing stock. Its indirect effects (prices, security of supply, compensation for projects already awarded) are not costed. Abolition of all support, a benchmark that is not an option of the cost schedule: projected 2027 charges according to the CRE (table 3 of deliberation 2026-149), onshore wind €1,404.6m, offshore wind €1,306.7m, photovoltaic €4,579.2m, electricity renewables in mainland France €8,378.5m in total, injected biomethane €1,109.8m in 2026 and €1,373.3m in 2027. These charges are not a static gain: breaking current contracts would give producers a right to compensation, of an order of magnitude close to the additional cost removed; supporting only future projects saves only the unsupported volumes (TWh parameter). A position that does not say which of the two it targets is not costed. Support for renewables in non-interconnected zones (overseas territories, Corsica) is not isolated by the CRE and is not counted. The amount follows the wholesale price: €2.9bn for electricity renewables in 2024, €8.4bn projected in 2027. The Fonds chaleur (heat fund) and tax expenditures are not counted. Cost of a new TWh: according to the CRE, the increase in supported volume from 88.8 to 98.5 TWh between 2026 and 2027, mainly photovoltaic, « génère une hausse de l'ordre de 860 M€ » [“generates an increase of around €860m”] (€0.089bn per TWh), and that from 2025 to 2026 (+14 TWh) about €1.3bn (0.093): the marginal cost of new volumes is close to the average additional cost, which limits the bias noted above. Volume targets (common rule of 18 October 2026): a target convertible into TWh (share of renewables in electricity consumption, end of aid for future installations) is costed by the gap, in supported TWh, with the PPE3 trajectory in its target year, maintained in 2032, multiplied by this coefficient; the editors take the PPE3 as current policy (photovoltaic: 48 GW and about 59 TWh in 2030, 55 to 80 GW in 2035, load factor of 14%; electricity renewables: about 205 TWh in 2030), and electricity consumption from the reference trajectory of RTE (the electricity transmission system operator) (451 TWh in 2025, 580 TWh in 2035, interpolated in a straight line; 505 TWh in 2035 in the slow decarbonisation trajectory). These conversions are editors' assumptions, flagged as such. A target with no milestone before 2032 or covering all energy (and not electricity) is not convertible and remains uncosted.

  • Délibération CRE n°2026-149 du 15 juillet 2026 (CSPE 2026-2027), 2026-07-15: « « TOTAL des charges à compenser en 2027 14 227,1 / Dont inscrites au budget de l'Etat 10 668,1 » (M€) ; « les surcoûts unitaires devraient se stabiliser respectivement à 89,4 €/MWh et 91,2 €/MWh » ; « le volume soutenu augmente de +10 TWh (98 TWh soutenus en 2027) » ; « l'augmentation du volume soutenu, de 88,8 TWh à 98,5 TWh, qui génère une hausse de l'ordre de 860 M€ » ; « entre 2025 et 2026, le volume soutenu augmente de +14 TWh (89 TWh soutenus en 2026), ce qui conduit à environ + 1,3 Md€ d'effet volume » ; « ont apporté 5,5 Mds€ de recettes au titre des années 2022 et 2023 » ; « Tableau 3 : Tableau récapitulatif des charges par action au titre des années 2024 à 2027 » ; colonne « Charges prévisionnelles au titre de 2027 » (M€) : « 1. Eolien terrestre » 1 404,6 ; « 2. Eolien en mer » 1 306,7 ; « 3. Photovoltaïque » 4 579,2 ; « 4. Bio-énergies » 855,1 ; « 5. Autres énergies » 232,9 ; « TOTAL » [soutien ENR électriques en métropole] 8 378,5 ; « 2. Injection biométhane » 1 373,3. Colonne « Mise à jour de la prévision au titre de 2026 » : TOTAL 7 286,6 ; injection biométhane 1 109,8 (cogénération et autres moyens thermiques : 653,1). Colonne « Charges constatées au titre de 2024 » : TOTAL 2 900,3. »
  • Programmation pluriannuelle de l'énergie 2026-2035, document annexé au décret n° 2026-76 du 12 février 2026 (Journal officiel du 13 février 2026, documents administratifs n° 3 ; copie en ligne sur environnementdurable.org), pages 27 et 91, 2026-02-13: « PHOTOVOLTAÏQUE […] 19,3 GW - 22,7 TWh 48 GW ~59 TWh 55 – 80 GW ~67 – 98 TWh […] Les facteurs de charges utilisés : 14 % pour le PV, 25 % en 2030, et 26 % en 2035 pour l'éolien terrestre et 45% pour l'éolien en mer. […] Jusqu'à 2028 : appels d'offres et arrêtés tarifaires dans la limite d'un plafond de 2,9 GW par an de puissance soutenue […] En 2030, la PPE devrait conduire à une production d'au moins 560TWh d'électricité décarbonée, avec au moins 200TWh (environ 205 TWh) d'électricité d'origine renouvelable »
  • CRE, rapport annuel 2026 à la Commission européenne (DG ENER), section 2.3.1.1 (consommation d'électricité), 2026-07-31: « La consommation d'électricité « à température normale » (corrigée des aléas météorologiques et calendaires), estimée par RTE, s'est élevée à 451 TWh en 2025 […] RTE prévoit une hausse de la consommation intérieure d'électricité de l'ordre de 30 % en 2035 par rapport à aujourd'hui (la consommation s'établissant à 580 TWh en 2035). […] Dans ce scénario, la consommation annuelle atteindrait 505 TWh en 2035 »
  • reseaux.photovoltaique.info (Hespul), « La PPE3 (à partir de 2026) », 2026: « A fin 2025, la puissance photovoltaïque installée est d'environ 31 GW, […] en 2030 : 59 TWh, soit 48 GW […] en 2035 : 67-98 TWh, soit 55-80 GW »
  • PPE3 : objectifs (synthèse de presse spécialisée, via résultat de recherche), 2026-02: « Photovoltaïque 48 GW en 2030 et 55 à 80 GW en 2035 ; éolien terrestre 31 GW en 2030 et 35 à 40 GW en 2035 ; éolien en mer 15 GW en 2035. »
Existing nuclear fleetnot costable
Current law, horizon, limitations and 1 source

Current law. Multiannual energy programme (PPE3, decree of February 2026): continued operation of existing reactors up to sixty years, or even beyond, under the supervision of the Autorité de sûreté nucléaire et de radioprotection (nuclear safety and radiation protection authority); no reactor closure scheduled.

Horizon. Not applicable for now.

Limitations. Not costed at this stage. A nuclear phase-out or an extension beyond sixty years mainly involves investment costs (grand carénage refurbishment programme, decommissioning, replacement capacity) and effects on the electricity price, borne largely by EDF and not by the State budget; no comparable public costing is cited.

Electricity prices, European marketcostable, revenue
  • Rate of the excise duty on electricity for sites of 36 kVA or less (households and small businesses). For €1/MWh of excise duty on the consumption of sites of 36 kVA or less (183 TWh in 2025), including at the central value the VAT on the excise duty paid by households: a gain of €0.212bn per year (0.183 to 0.22) (coefficient calculated by the editors from CRE, rapport annuel 2026 à la Commission européenne (figures 22 et 23)). Current value: 30.62 €/MWh. Range measured by the source: 0 to 30.62 €/MWh; beyond it, linear extrapolation is flagged. Coefficient proportional to the taxed volume (editors' calculation): arithmetic range from abolition of this rate (€0/MWh) to its current level. An increase beyond that is a linear extrapolation, flagged as such. Rates for sites above 36 kVA are not affected by this parameter.
  • 10% cut in the excise duty on electricity (all tariffs): a cost of €0.82bn per year (0.8 to 1).
  • Abolition of the excise duty on electricity: a cost of €8.2bn per year (8.2 to 10).
Current law, horizon, limitations and 8 sources

Current law. France remains in the integrated European electricity market. ARENH ended on 31 December 2025. It is replaced by the universal nuclear payment (versement nucléaire universel, art. 17 of the LF 2025, act no. 2025-127): half of EDF's revenue above €78/MWh is paid back, and 90% above €110/MWh (thresholds set by the order of 11 February 2026; rates of 50% and 90% read in summaries, not in the text). The VNU is not expected to be triggered in 2026 (reference price estimated at around €66/MWh). Regulated sales tariffs are maintained. The excise duty on electricity for households (36 kVA or less) has been €30.62/MWh since 1 August 2026, including the ZNI surcharge. It yielded €8.2bn in 2025.

Horizon. Revenue shortfall in a full year, static, 2025 base (excise duty) or 2024 base (VAT).

Limitations. Leaving the European market has no measurable direct budgetary cost: it is a matter of European law and of EDF's rent (lower dividends, effects on the VNU, universal nuclear payment), which cannot be costed statically. The value 0 means “not costed”, not “free”. Excise duty: €8.2bn is the 2025 yield according to a CRE note of September 2026, taken up by a comparison website. The upper bound of €10bn comes from a projection for 2026, not official. Tariffs differ by consumer category, so a decrease of X €/MWh does not convert exactly. VAT: the Institut Montaigne costing also covers fuels, and the EU does not allow the reduced rate on fuels in mainland France. This sub-lever intersects with the “Taxes on fuel and energy” lever in domain C: it is assigned to a single lever to avoid double counting. VAT at 5.5% on energy is under the « tva » (VAT) lever (once only); leaving the European market has no measurable static budgetary cost. Target prices (editors' decision of 18 October 2026, the same for all candidates): any target price for electricity or gas (€75/MWh, €100/MWh, a 30% decrease, prices based on production costs, contract for difference, tariff freeze) is not costed, with the same reason, word for word: “Target price: no budgetary cost as long as the gap with the market price is not offset by the State; if it is, it becomes spending (benchmark: the 2022-2024 tariff shield, costed by the Cour des comptes).” Only tax measures are costed (the excise duty here, VAT under the « tva » (VAT) lever), on the same base for all. Excise duty rate for sites of 36 kVA or less, coefficient calculated by the editors: 146.4 TWh consumed in 2025 by residential sites and 37 TWh by small non-residential sites, i.e. 183 TWh (CRE, 2025 data); €1/MWh is worth €0.183bn excluding VAT (lower bound). Central value, 0.212: the 20% VAT on the excise duty paid by households (146.4 TWh), i.e. 0.2 × 0.146 ≈ 0.029 more; that paid by businesses is recoverable. Upper bound, 0.220: VAT counted on the 183 TWh. Same convention as for the excise duty on fuels. Consumption follows the weather and electrification: the coefficient is not updated to 2032. Tariff benchmarks: an order published in the Journal officiel of 31 January 2024 raised, from 1 February 2024, the rates lowered by the tariff shield from €1 to €21/MWh for households and from €0.5 to €20.5/MWh for businesses (+€20/MWh); the excise duty on natural gas for households has been €16.66/MWh since 1 August 2026 (€16.39/MWh from 1 February to 31 July 2026). A convergence of excise duties on gas and electricity can be budget-neutral: the version voted by the Senate in November 2025 (−€3/MWh on household electricity, +€4.40/MWh on gas) was presented as such; it was not retained in the finance act.

Agriculture and trade

Floor prices, farm incomenot costable
Current law, horizon, limitations

Current law. The EGalim acts (2018, 2021 and 2023) provide for price-setting « en marche avant » [“moving forward”, i.e. starting from the producer], production cost indicators and the non-negotiability of agricultural raw materials in certain contracts. There is no general statutory floor price. These texts are not cited in the source (Légifrance, vie-publique and the DGCCRF were not accessible); references: act no. 2018-938 and act no. 2021-1357.

Horizon. Not applicable.

Limitations. A floor price is a pricing rule between private actors. It has no direct budgetary cost, but an effect on food prices that no official source quantifies. If a proposal runs through public aid (income support, State-funded price guarantee), it is costed separately, at the announced amount.

Pesticidesnot costable
Current law, horizon, limitations and 2 sources

Current law. Act no. 2025-794 of 11 August 2025 aiming to lift constraints on farming (loi Duplomb). The Constitutional Council (decision no. 2025-891 DC of 7 August 2025) struck down the reintroduction by derogation of acetamiprid, a neonicotinoid, on the basis of the Charte de l'environnement (Environment Charter). The Ecophyto 2030 strategy sets targets for reducing risks and use; the percentage and the indicator are not cited in the source.

Horizon. Not applicable.

Limitations. This is a regulatory lever, with no significant budgetary cost. The support plans (Ecophyto, transition aid) have announced amounts, used as they are if a candidate changes them. The quantified targets of Ecophyto 2030 are not cited from the source (agriculture.gouv.fr pages returned a 404 error during the research).

Common Agricultural Policynot costable
Current law, horizon, limitations and 2 sources

Current law. The CAP national strategic plan 2023-2027 allocates French agriculture about €9.1bn a year (official document of 2022). Secondary sources give €9.4bn a year, of which €8.7bn for the PSN. The transfer from the first to the second pillar is 7.5%. The post-2027 EU financial framework is under negotiation.

Horizon. Not applicable.

Limitations. The PAC (Common Agricultural Policy) is funded by the EU budget. A national position (capping, convergence, greening) redistributes the envelope with no net cost for the national budget. Only national co-financing of the second pillar or a substitute national aid would have a cost. If a candidate proposes to « renationaliser » [“renationalise”] the PAC, the cost would be about €9bn per year of national spending, less the reduction in the French contribution to the EU budget, which is not costed here.

Free trade agreements (Mercosur)not costable
Current law, horizon, limitations and 2 sources

Current law. EU-Mercosur agreement: the Council authorised its signature on 9 January 2026 by qualified majority, despite France voting against. On 21 January 2026, the European Parliament referred it to the CJEU for an opinion (334 votes for, 324 against, 11 abstentions), which suspends its consent. The Commission announced provisional application of the interim trade agreement from early May 2026; its actual entry into application has not been checked. Trade policy is an exclusive competence of the EU.

Horizon. Not applicable.

Limitations. No measurable direct budgetary effect. Customs duties are an EU own resource. Sectoral effects (agriculture, industry) are a matter for impact studies, outside the static costing. A national « sortie » [“exit”] or « refus » [“refusal”] has legal force only for the part of the agreement subject to national ratification.

Industry and investment

Additional public investment, per yearcostable, spending
  • Additional public investment. For +€1bn of public investment per year: a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
Current law, horizon, limitations and 1 source

Current law. There is no national rule setting the level of public investment. The 2024 European fiscal rules (net expenditure path) apply. General government GFCF amounts to €132.2bn in 2025 (Insee, annual national accounts, base 2020), i.e. about 4.4% of a GDP of €2,991.1bn.

Horizon. Spending in a full year, static: €1 announced = €1 of deficit, before any feedback effect on growth.

Limitations. Most public investment is carried out by local authorities: a State commitment is only met if it goes through grants or State appropriations. Beware of double counting with the rail, renovation, EPR2 (a loan is not public investment), France 2030 and defence levers. A plan that goes partly through loans, guarantees or capital injections weighs less on the deficit: the Institut Montaigne (2022) estimated, « à la lumière » [“in the light”] of the investment-for-the-future programmes, that €200bn over five years would weigh only €29bn per year on the deficit (72.5%). These financial transactions weigh on the debt, which the cost schedule counts separately; the parameter applies only to investment spending, and this judgement on the composition of a plan is not a bound of the coefficient. Multiplier effects are not included in the central figure (rule of the method). Provisional Insee data, subject to revision. Common rules of 18 October 2026: an announced envelope is taken at the candidate's amount, converted to an annual figure over the duration they give or, if they give none, over the five-year term (five years); a fund lent by the Caisse des dépôts, Bpifrance or regulated savings (public banking hub) is read as having no effect on the balance (low reading: position counted at zero and marked as an assumption), the reading as public spending being cited in the note.

  • Comptes nationaux annuels en 2025 (Insee), 2026-05: « FBCF « administrations publiques 5,4 5,3 2,6 132,2 1,0 3,7 0,1 » (132,2 = milliards d'euros 2025) ; le PIB « atteint 2 991,1 milliards d'euros » »
Industrial aid and reshoringcostable, spending
  • Additional public aid to industry. For +€1bn of public aid to industry per year (subsidies or tax expenditure): a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
  • Extend France 2030 beyond 2030 at the pace of 2026 appropriations (vs the plan's expiry): a cost of €5.1bn per year (4 to 6).
Current law, horizon, limitations and 2 sources

Current law. France 2030 plan: €54bn (including PIA 4). Payment appropriations proposed in the PLF 2026 for its two programmes amount to €5,128m. The end of roll-out was envisaged around 2027-2030. The green industries investment tax credit (C3IV) is extended for projects approved until 31 December 2028 (art. 39 of the LF 2026, act no. 2026-103; decree no. 2026-763). The rate was cut by 5 points according to two analyses, which has not been checked in the text.

Horizon. In full year 2032, compared with current policies under which France 2030 has expired.

Limitations. The figure of €5.1bn corresponds to the payment appropriations (CP) proposed in the PLF 2026 and not voted (the Senate cut them by €1bn at first reading). The bounds of €4bn to €6bn are an order of magnitude chosen by the editors around this figure, not a source. The annual cost of the C3IV (green industry investment tax credit) is not in the sources read (« Voies et moyens » annex, volume II, of the PLF 2027, not consulted). A 2025 amendment puts forward about €3.7bn cumulatively up to 2030. A “conditionality of aid” costs nothing and can save money if it reduces the amounts. Compatibility with EU State aid law is not assessed.

Nationalisations and privatisationscostable, spending
  • Buybacks of shareholdings (annual interest charge). For €10bn of shareholding buybacks (additional debt), annual interest charge: a cost of €0.3bn per year (0.27 to 0.31) (coefficient calculated by the editors from Agence France Trésor — Rapport annuel 2025 / Bulletin mensuel juillet 2026 (via résultat de recherche)). Current value: 0 €bn. Range measured by the source: 0 to 50 €bn; beyond it, linear extrapolation is flagged. Interest charge at the AFT's issuance rate (editors' calculation): range up to the largest buyback costed by a source read (motorways, €45bn to €50bn). Beyond that, linear extrapolation, flagged as such.
  • Early buy-back of the motorway companies (€45bn-€50bn) — gross interest charge: a cost of €1.4bn per year (1.2 to 1.6) (coefficient calculated by the editors from Agence France Trésor — Rapport annuel 2025 / Bulletin mensuel juillet 2026 (via résultat de recherche)).
Current law, horizon, limitations and 3 sources

Current law. The State has held 100% of EDF since the 2023 takeover bid (buyout of the 15.9% of the capital and of the convertible bonds it did not hold, about €9.7bn according to the Ministry of the Economy). The historic motorways are under concession until 2031-2036. There is no privatisation programme under way according to the sources read.

Horizon. Interest charge in a full year, at the AFT issuance rate (2.73% on average in 2025, 3.14% for long-term rates at end-2025). The purchase price is a financial transaction: it increases the debt, not the deficit.

Limitations. The figure does not take into account dividends received or lost, which may offset or exceed the interest charge. For motorways, the Senate commission of inquiry (2020) estimated dividends at about €40bn beyond 2022 until expiry. The price of a nationalisation depends on the premium and on litigation: the Cour des comptes (May 2026, read in a blog article) reportedly judged that a lower premium on EDF would have saved €1bn. Sign convention for options: negative = gain.

Transport

Everyday transportcostable, spending
  • Entirely free urban public transport throughout France (lost fare revenue, 2019 level): a cost of €5.7bn per year.
  • Entirely free urban public transport outside Île-de-France (lost fare revenue, 2019 level): a cost of €1.5bn per year.
  • Entirely free public transport in Île-de-France (lost fare revenue, 2017 level): a cost of €3.6bn per year.
  • Completely free urban public transport in the 28 provincial networks with a metro or tramway (fare revenue lost, 2019 level): a cost of €1.254bn per year.
Current law, horizon, limitations and 5 sources

Current law. Mobility organising authorities (intermunicipal bodies, regions, Île-de-France Mobilités) define pricing policy and set fares (Code des transports, art. L. 1221-5). Employers cover half of their employees' public transport season ticket (Code du travail, art. L. 3261-2; 75% for public employees since 1 September 2023), i.e. €1,284m in 2023, and those with 11 or more employees pay the versement mobilité (mobility levy) (€11.3bn in 2023). In 2019, fare revenue covered 41% of the operating expenditure of urban public transport, about €5.7bn out of €13.9bn; 46 areas offer entirely free transport (Cour des comptes, September 2025). Metropolitan regional express services (SERM) have had a legal framework since the act of 27 December 2023.

Horizon. Full year, €bn per year, in 2019 euros (2017 for Île-de-France alone), not updated to the 2032 horizon. Cost for general government as a whole, whichever authority bears it.

Limitations. Static costing: free public transport costs first of all the lost fare revenue, offset by the budget of the transport authorities (classified here as spending). Not counted: the additional service that higher ridership would require (the Cour des comptes considers free transport risky for large networks), savings on ticketing and inspection (€250m at best in Île-de-France according to the Senate), or the saving made by employers on their contribution to season tickets (€1,284m in 2023, included in fare revenue); shifting this share onto the versement mobilité (mobility levy) is a means of funding, not a reduction in cost. The first three options do not add up: all of France = Île-de-France + the rest of France, for different years (€5.7bn in 2019 versus 3.6 + 1.5 = €5.1bn). The fourth, the 28 networks outside Île-de-France with a metro or tramway (€1,254m of revenue in 2019 out of €1,638m outside Île-de-France, Cour des comptes, table no. 1), is the best approximation read for the networks of the métropoles outside Île-de-France: it counts a few conurbations that are not métropoles and leaves out métropoles with neither metro nor tramway; it is included in the rest of France and does not add to the other options. Outside the scope for lack of a source read: TER and regional trains, intercity coaches, partial free transport (young people, weekends; Lille and Nantes give local examples at €8m and €16m per year), fare reductions, rail pass, metropolitan RER (major projects are costed at the announced amount).

Railcostable, spending
  • Additional public rail investment. For +€1bn a year of public rail investment: a cost of €1bn per year (candidate’s amount, taken as is). Current value: 0 €bn/year.
  • Meet the identified renewal need (+€1.5bn/year; €2bn including inland waterways according to the Senate): a cost of €1.5bn per year (0 to 2).
Current law, horizon, limitations and 2 sources

Current law. Renewal and modernisation of the rail network amount to about €3bn a year, stable, financed by SNCF Réseau (own resources and contributions from SNCF Voyageurs), not by the State budget (Senate, opinion a25-143-2). The « Ambition France Transports » conference (9 July 2025) assesses the need at €1.5bn a year more; « à partir de 2028 » [“from 2028”] and the total effort of €4.5bn a year come from the press (Banque des territoires), to be financed in time by the future motorway concessions. The framework transport bill (tabled on 11 February 2026) was adopted in the Senate on 28 April 2026 and was in committee at the National Assembly in July 2026: it has not been finally passed. The resources of AFIT France planned in the PLF 2026 amount to €3.75bn.

Horizon. Public spending in full year 2032, compared with current policies under which the 2028 commitment is not written into an enacted law.

Limitations. The lower bound of 0 corresponds to the case where the government commitment « dès 2028 » [“from 2028”] would be a current policy; it is not written into a law that has been voted, and the central value counts it as a cost, with the same rule for all candidates. Part of the effort is borne by SNCF Réseau (attributed revenue (fonds de concours), SNCF group profits), so not everything goes through the State budget. Major projects (LGV high-speed lines, metropolitan RER, freight with the Ulysse Fret plan of about €4.5bn) are costed at the announced amount. A total envelope with no duration is converted to an annual figure over the five-year term (five years), common rule of 18 October 2026 (see public investment).

Motorway concessionscostable, spending
  • Early buy-back / nationalisation before expiry (€45bn-€50bn) — gross interest charge, excluding captured dividends: a cost of €1.4bn per year (1.2 to 1.6) (coefficient calculated by the editors from Sénat, rapport n° 709 (2019-2020) de la commission d'enquête sur le contrôle, la régulation et l'évolution des concessions autoroutières (pages 17 et 18 du PDF)).
  • Direct public operation at expiry (2031-2036) instead of new concessions — static, compared with the framework bill: a cost of €0bn per year.
  • 10% cut in tolls under direct public operation: a cost of €1.3bn per year (1.2 to 1.3) (coefficient calculated by the editors from Sénat, rapport n° 709 (2019-2020) de la commission d'enquête sur le contrôle, la régulation et l'évolution des concessions autoroutières (pages 17 et 18 du PDF)).
Current law, horizon, limitations and 4 sources

Current law. The historic concessions (more than 90% of the conceded network) expire between 2031 and 2036. The framework transport bill, adopted in the Senate and not yet finally passed, provides for new concessions of which about €2.5bn a year in revenue would be allocated to transport from 2032. The 2024 turnover of the concession companies is €12.8bn (ART). Since the LFI 2024, a tax on the operation of long-distance transport infrastructure has been allocated to AFIT up to €500m.

Horizon. Full year 2032. The buy-back is a financial transaction: it increases the debt by €45bn to €50bn once, and the deficit only through interest.

Limitations. Buyback: same measure as the “motorway buyback” option of the nationalisations lever (overlap declared); interest is calculated at AFT rates of 2.73% to 3.14%. Profits captured by the State would have to be deducted: the Senate estimated dividends at about €40bn between 2022 and 2036, but dividends paid to shareholders do not become net public revenue after a buyback (corporate income tax, investment). The net figure may therefore become a gain, with no official costing. The Senate estimate dates from 2020 and mechanically moves closer to zero as expiry approaches. Direct public management at expiry: the assets revert to the State free of charge. Compared with the framework bill, which provides about €2.5bn per year for transport, the static balance is close to zero but depends on the arrangement. Tolls: 10% × €12.8bn (2024 turnover, ART summary).

Institutions, sovereignty, society

Institutions

Voting system for legislative electionsnot costable
Current law, horizon, limitations and 4 sources

Current law. At most 577 MPs (Constitution, art. 24 para. 3), elected by single-member two-round majority vote in each constituency (Code électoral, art. L123). This voting system falls under ordinary law: proportional representation (full or partial) does not require a revision of the Constitution.

Horizon. Not applicable.

Limitations. No budgetary effect of its own: changing the voting system does not alter the cost of an election (€171.49m for the 2024 legislative elections under programme 232, €164.47m for those of 2022, given as an order of magnitude, useful only if a measure adds a vote).

Citizens' initiative referendum (RIC)costable, spending
  • National referendums held per year. For 1 national referendum held (cost of one vote): a cost of €0.15bn per year (0.1 to 0.2). Current value: 0 elections. Range measured by the source: 0 to 1 elections; beyond it, linear extrapolation is flagged. Cost of one vote according to the Minister of the Interior (2025); for several votes, linear extrapolation, flagged as such.
  • RIC (citizens' initiative referendum) with votes grouped with elections already scheduled: a cost of €0.05bn per year (0 to 0.1).
Current law, horizon, limitations and 3 sources

Current law. No RIC (citizens' initiative referendum). There is only the shared-initiative referendum (Constitution, art. 11 para. 3, 2008 revision): it requires the initiative of one fifth of parliamentarians, supported by one tenth of registered voters. A national referendum is held on the initiative of the President (art. 11) or as part of a revision (art. 89).

Horizon. Cost per vote, in current €bn. The annual cost in 2032 depends on the number of referendums, which the programme must specify.

Limitations. Minimal budgetary effect compared with the democratic stakes. Bottom of the range: an electronic vote, according to the minister. Top: a paper vote. The effects of decisions taken by RIC (abolition of taxes, etc.) are costed in the relevant levers, not here. The cost of the 2005 referendum (€130.6m in 2005 euros, borne by the Ministry of the Interior alone) comes from the ministerial reply; it is within the €100m to €200m range given by the minister in 2025. Nearly half (€58.8m) is the sending of documents to voters.

Article 49.3 and revision of the Constitutionnot costable
Current law, horizon, limitations and 1 source

Current law. Article 49 para. 3 of the Constitution: the Prime Minister may, after deliberation by the Council of Ministers, make the passing of a text an issue of the Government's responsibility. Since the 2008 revision, this is possible only for finance bills and social security financing bills, plus one other text per session. Revision of the Constitution: article 89, either by referendum or by Congress with a 3/5 majority.

Horizon. Not applicable.

Limitations. Procedural lever, with no direct budgetary effect. A revision adopted by referendum costs the price of a vote: see the ric lever (€0.1bn to €0.2bn per vote).

  • Constitution, texte en vigueur (Conseil constitutionnel): « Art. 49 al. 3 : « Le Premier ministre peut, après délibération du conseil des ministres, engager la responsabilité du Gouvernement » ; art. 89 al. 3 : « Toutefois, le projet de révision n'est pas présenté au référendum lorsque le Président de la République décide » »
Length and number of terms of officenot costable
Current law, horizon, limitations and 1 source

Current law. President elected for five years, at most two consecutive terms (Constitution, art. 6, 2000 reform and 2008 revision). MPs elected for five years. Holding a parliamentary seat together with a local executive office has been prohibited since 2017 (organic law no. 2014-125). Senate: at most 348 members (art. 24).

Horizon. Not applicable.

Limitations. Lengthening or shortening terms of office changes the number of elections only marginally. A reduction in the number of parliamentarians would have a budgetary effect, but no grant to the assemblies is recorded here. The organic law on multiple office-holding is cited without a link to the text.

  • Constitution, texte en vigueur (Conseil constitutionnel): « Art. 6 : « Le Président de la République est élu pour cinq ans au suffrage universel direct. » « Nul ne peut exercer plus de deux mandats consécutifs. » ; art. 24 al. 4 : « Le Sénat, dont le nombre de membres ne peut excéder trois cent quarante-huit » »
Voting agenot costable
Current law, horizon, limitations and 2 sources

Current law. Voters are French nationals of full age (Constitution, art. 3) who have reached eighteen (Code électoral, art. L2, as worded by act no. 74-631 of 5 July 1974, which lowered the age from 21 to 18). The age of civil majority is set at eighteen (Code civil, art. 414). Lowering the voting age to 16 for all elections would require, on the usual reading, revising article 3 of the Constitution or lowering the age of majority.

Horizon. Not applicable.

Limitations. Electoral rule, with no direct static budgetary cost: cannot be costed (the cost of registering about two additional age cohorts on the electoral rolls is negligible and has not been costed by any source). Article 414 of the Civil Code and the 1974 law are cited without online checking; the need for a constitutional revision is a doctrinal reading, not a decision.

Immigration

Labour immigration and residence permitsnot costable
Current law, horizon, limitations and 2 sources

Current law. Act no. 2024-42 of 26 January 2024: exceptional admission to residence for foreign nationals working in a shortage occupation (CESEDA, art. L. 435-4). It requires three years of residence and 12 months of work out of 24, and the scheme is open until 31 December 2026. In 2025, 384,230 first residence permits were issued (+11%), of which 51,190 for economic reasons (-13%), and 28,610 foreign nationals were regularised (-10%).

Horizon. Not applicable.

Limitations. A static costing makes no sense: the budgetary effect runs through employment, contributions and benefits, that is, through dynamic effects. Statistics diverge: about 377,000 first permits according to the ministry's note, 384,230 according to AFP (scope with or without British nationals). Expiry date of 31 December 2026 read in the instruction of 5 February 2024: without an extension (not searched for), the « métiers en tension » (shortage occupations) scheme expires before 2027, which changes the reference law in 2032.

Right of asylumcostable, spending
  • Change in the number of asylum applications registered per year (negative when it falls). For 10,000 fewer asylum applications per year (material support for applicants, average cost): a gain of €0.09bn per year (0.016 to 0.09) (coefficient calculated by the editors from Sénat, commission des finances, note de présentation PLF 2026, mission « Immigration, asile et intégration »). Current value: 0 applications. Range measured by the source: -151,665 to 151,665 applications; beyond it, linear extrapolation is flagged. Average cost per application (151,665 applications in 2025, editors' calculation): arithmetic range from no applications at all (−151,665) to their doubling (+151,665); each decrease of 10,000 applications is a gain, each increase a cost (negative step, changed on 18 October 2026 so that the direction follows the change in applications). Beyond that, linear extrapolation, flagged as such.
  • Abolition of the asylum seekers' allowance (excluding temporary protection), contrary to the EU « accueil » [“reception”] directive: a gain of €0.24bn per year.
Current law, horizon, limitations and 5 sources

Current law. Right of asylum guaranteed by the CESEDA and EU law. The European Pact on Migration and Asylum, adopted on 14 May 2024, applies from June 2026. In 2026, the « Immigration, asile et intégration » mission receives €2.16bn in payment appropriations, 64% of it for asylum (asylum seeker allowance and accommodation). Asylum seeker allowance: €313.4m in 2026, of which €241.4m for applicants. Accommodation for asylum seekers: 111,855 places budgeted in 2026. Asylum applications: 151,665 in 2025 (AFP), 145,270 lodged with the Ofpra (French refugee protection office) (2025 activity report), of which 8,897 in Guyane and 1,593 in Mayotte, and 1,677 requests for entry on asylum grounds made at the border. Cour nationale du droit d'asile (National Court of Asylum): 53,086 decisions in 2025, of which 10,413 by a single judge after a hearing (1,469 single-judge hearings), since the Act of 26 January 2024.

Horizon. Full year, current €bn, based on 2026 appropriations (no projection to 2032 in the sources)

Limitations. The unit cost is calculated from the sources: 64% × €2.16bn ≈ €1.38bn, divided by 151,665 applications, i.e. about €9,100 per application and €0.09bn for 10,000 applications. Bottom of the range: the asylum seekers' allowance alone (€241.4m / 151,665 × 10,000 ≈ €0.016bn), if the accommodation stock remains fixed in the short term. No source gives a top of the range, hence top = central. Not counted: costs of OFPRA and CNDA, of AME for rejected applicants, of mainstream emergency accommodation. The mission budget covers only part of the cost of immigration: €7.82bn across 12 missions, and €1.8bn per year for the fight against irregular immigration alone according to the Cour des comptes (2024). The “abolition of the allowance” option is legally contrary to EU law, hence a purely indicative costing. Applications registered on French territory, for a measure that only accepts applications made at the border or outside France: 151,665 − 1,677 ≈ 150,000 (editors' calculation). Return to collegial panels at the CNDA (editors' calculation, not retained as an option: under €0.01bn, it would show as a zero cost): 1,469 single-judge hearings in 2025 (CNDA) × two assessors × €145 to €195 per sitting (Order of 12 December 2018, art. 3; amounts possibly amended by the Order of 31 December 2024, not read), i.e. €0.4m to €0.6m per year; the presiding judge of the panel is already the single judge; travel and premises not counted.

State medical aid (Aide médicale de l'État)costable, spending
  • Total abolition of the AME (state medical aid) with no replacement scheme (gross saving = projected spending in 2032): a gain of €1.82bn per year (1.7 to 2.08) (coefficient calculated by the editors from Sénat, commission des affaires sociales, L'Essentiel PLF 2026 « Santé » (avis Florence Lassarade)).
  • Transformation into emergency medical aid (limited basket: emergencies, serious illnesses, pregnancy, vaccinations): a gain of €0.4bn per year (0.2 to 0.7) (coefficient calculated by the editors from Sénat, commission des affaires sociales, L'Essentiel PLF 2026 « Santé » (avis Florence Lassarade)).
  • Prior approval extended to all scheduled non-urgent care (amendment by the Senate social affairs committee to the PLF 2026): a gain of €0.2bn per year (0.1 to 0.2).
Current law, horizon, limitations and 2 sources

Current law. AME (state medical aid; Code de l'action sociale et des familles, art. L251-1 et seq.): coverage of healthcare for irregular migrants resident in France for more than 3 months, subject to an income ceiling. Prior approval for certain scheduled care during the first 9 months. Actual spending of €1,386.8m in 2024 (+15.5%) and €1,385m forecast for 2025 (DSS), whereas the appropriations in the PLF 2026 are €1,208.3m. 461,833 beneficiaries at 31 March 2025. Spending split 61% hospital care and 39% community care.

Horizon. Full year 2032, current €bn, compared with current law

Limitations. 2032 projection calculated by the editors: €1,385m in 2025 × (1+g)^7, with g = 3% (low), 4% (central, rate anticipated by the Senate for 2026) and 6% (high, average since 2009). Abolition: GROSS and static saving. It ignores the shift to urgent care (€132m in 2024, already partly borne by national health insurance), hospitals' unrecoverable debts and health effects. The Évin-Stefanini report (Dec. 2023) considers the AMU (emergency medical aid) « inappropriée » [“inappropriate”] and costs no saving. There is no official costing of the AMU. The range is constructed: the low end takes the order of magnitude of the Senate amendment (-€200m), the high end removes the whole share of outpatient care (39% × 1.82 ≈ €0.7bn), the central value is halfway. Voted appropriations underestimate actual spending every year: the State's debt to social security in respect of AME (state medical aid) would reach €362m to €428m at the end of 2025. The cost schedule therefore costs on actual spending, not on appropriations. The three options are mutually exclusive (overlaps declared).

  • Sénat, commission des affaires sociales, L'Essentiel PLF 2026 « Santé » (avis Florence Lassarade), 2025-12-03: « « Les dépenses d'AME ont progressé de 6 % par an en moyenne depuis 2009, et de 40 % entre 2021 et 2025. Alors que les crédits consacrés à l'AME resteraient plafonnés à 1 208,3 millions d'euros en 2026, les dépenses réelles d'AME ont atteint 1 387 millions d'euros en 2024 » ; « La direction de la sécurité sociale (DSS) estime ainsi que la dépense prévisionnelle s'établira à 1 385 millions d'euros pour l'exercice 2025 » ; « En 2026, une progression de 4 % est anticipée » ; « Les prestations hospitalières représentent 61 % de cette dépense » ; « un amendement minorant les crédits de l'AME de 200 millions d'euros » »
  • Sénat, rapport d'information n° 841 (2024-2025), Vincent Delahaye, commission des finances, sur l'aide médicale de l'État, 2025-07: « « le coût total réel de l'AME est de 1 386,8 millions d'euros, soit une hausse de 15,5 % des dépenses par rapport à 2023 » ; « l'État ne prend pas en charge l'intégralité de la dépense d'AME pour soins urgents, puisqu'il ne verse que 70 millions d'euros à la sécurité sociale sur un coût total de 132 millions d'euros » »
Birthright citizenship and nationalitynot costable
Current law, horizon, limitations and 3 sources

Current law. Code civil: double jus soli (art. 19-3, child born in France to a parent who was also born in France) and acquisition of nationality at majority for a child born in France to foreign parents, with residence (art. 21-7). Special regime in Mayotte: since the act of 12 May 2025, both parents must have resided lawfully in France for at least one year at the time of birth (only one if parentage is established only with respect to that parent). The Constitutional Council upheld it with a reservation, and the implementing texts were published in May 2026.

Horizon. Not applicable.

Limitations. No static budgetary effect. Abolishing the droit du sol (birthright citizenship) nationally would probably require a revision of the Constitution: the Constitutional Council accepted the Mayotte regime only because of the « spécificités » [“specific characteristics”] of that territory. Articles 19-3 and 21-7 of the Civil Code are cited without a link to the text (Légifrance was not accessible during the research). The date of the law differs between sources (12 or 13 May 2025: date of the law or of its publication in the JO, the official journal).

Europe and defence

Place in the European Unioncostable, spending
  • Levy on revenue for the European Union (gross French contribution). For €1bn less in French contribution compared with the levy voted for 2026 (€28.44bn, Finance Act for 2026, art. 143): a gain of €1bn per year. Current value: 28.44 €bn. Range measured by the source: 0 to 38 €bn; beyond it, linear extrapolation is flagged. Gain equal to the difference from the levy voted for 2026 (€28.44bn): range from abolition to the average proposed by the Commission for 2028-2034 (€38bn). A targeted net contribution is converted into the gross levy by adding the average returns (€16.4bn), assumed unchanged.
  • Freeze of the French contribution at its 2025 level (amendment no. 1291 to the PLF 2026): a gain of €5.34bn per year.
  • Leaving the EU: static budgetary gain equal to the net balance (excluding any macroeconomic effect): a gain of €12.04bn per year (7.9 to 21.6) (coefficient calculated by the editors from Sénat, commission des finances, L'Essentiel PLF 2026, mission « Affaires européennes » (art. 45, prélèvement sur recettes UE, J.-M. Mizzon)).
Current law, horizon, limitations and 6 sources

Current law. Constitution, art. 88-1 (participation in the EU); TEU and TFEU treaties; new European fiscal rules of 2024. Levy on revenue for the EU: €28.44bn in the finance act for 2026 (art. 143; €28.8bn in the initial bill), compared with €23.1bn in 2025. France's share of EU financing: 17.2% in 2024. Net accounting balance: -€7.9bn in 2024 and -€9.3bn in 2023, France being the 2nd largest net contributor. Returns: €16.4bn a year on average. Under the 2028-2034 financial framework proposed by the Commission, the average annual levy would rise from €26bn (current framework) to €38bn.

Horizon. Full year, current €bn. In 2032, the 2028-2034 financial framework brings the contribution to €38bn on average according to the Commission's proposal.

Limitations. The contribution is set by the « ressources propres » (own resources) decision (unanimity and ratification) and by the multiannual financial framework: a unilateral reduction would breach the treaties. The “exit” option is a static balance calculated by the editors. Low: 2024 balance (7.9). Central: voted 2026 levy of 28.44 minus 16.4 of average returns, i.e. 12.04. High: 38 (average proposed for 2028-2034) minus 16.4, i.e. 21.6. It ignores the returns that would have to be funded nationally (PAC: France receives 18.5% of the agricultural policy), customs duties, access to the single market, the cost of debt and the French share of the European debt linked to the recovery plan (extra contribution of about €2.5bn per year from 2028 according to the Cour des comptes, for lack of new own resources). It is not a net gain. The levy voted for 2026 is €28.44bn (finance act for 2026, art. 143), not €28.8bn (initial bill). Freezing at the 2025 level brings in 28.44 − 23.10 = €5.34bn according to amendment no. 1291 (GDR group), which proposes it. The contribution-ue parameter (rule on public bases, 18 October 2026, the same for all): a reduction in the contribution is costed on the levy voted for 2026, €28.44bn, assuming the reduction is obtained, as for leaving the EU (static costing; the means, a renegotiation requiring unanimity, is not assessed). A targeted net contribution (levy minus returns) is converted into a levy by adding the average returns, €16.4bn (Senate): a net contribution of €5bn is equivalent to a levy of €21.4bn, i.e. a gain of €7.04bn (only €2.9bn if one starts from the 2024 net accounting balance, €7.9bn). A return to the pre-Covid level is read on the levy executed in 2019, €21.025bn, in 2019 euros not updated: a gain of €7.41bn. Compared with the proposed 2028-2034 framework (€38bn on average), the gains would be higher, but this framework has not been adopted: it is not part of the baseline.

Defence budgetcostable, spending
  • Defence effort in 2032. For +0.1 point of GDP of defence effort in 2032: a cost of €3.68bn per year (3.56 to 3.81) (coefficient calculated by the editors from Légifrance, loi n° 2026-791 du 16 août 2026 actualisant la programmation militaire pour les années 2024 à 2030). Current value: 2.5 % of GDP. Range measured by the source: 2 to 3.5 % of GDP; beyond it, linear extrapolation is flagged. Range set by the military programming act: from 2% of GDP (2025) to the 3.5% target (2035). Beyond that, linear extrapolation, flagged as such.
Current law, horizon, limitations and 6 sources

Current law. Act no. 2026-791 of 16 August 2026 updating the 2024-2030 military programming: national defence effort raised from 2% of GDP in 2025 « à 2,5 % au minimum en 2030 » [“to at least 2.5% in 2030”], with the aim of reaching 3.5% by 2035. Payment appropriations for the « Défense » mission: €57.1bn (2026), 63.3 (2027), 69.5 (2028), 72.6 (2029), 75.7 (2030), i.e. €435.7bn over 2024-2030 (+€36bn over 2026-2030). NATO Hague commitment (June 2025): 5% of GDP in 2035, of which at least 3.5% on defence in the strict sense, with a review in 2029.

Horizon. Full year 2032, current €bn, static. 2032 GDP calculated by the editors: 2025 GDP (€2,991.1bn) × nominal growth of 2.5% / 3% / 3.5% per year over 7 years, i.e. €3,556bn / €3,679bn / €3,806bn.

Limitations. Cost schedule rule: the reference law is the latest level set by the military programming law in force, at least 2.5% of GDP in 2030, maintained in 2032 (current value of the parameter: 2.5). Beyond 2030, the law sets only a TARGET of 3.5% in 2035, without annual steps: it is not part of the baseline. A position at 3.5% or at 2% is costed by the linear parameter from this baseline; a position “at least 3%” is costed at the 3% floor. Nominal growth of 2.5% to 3.5% per year up to 2032 is a calculation assumption, not an official forecast. The « effort national de défense » (national defence effort) in the NATO sense includes military pensions and differs from the appropriations of the « Défense » mission (€75.7bn in 2030 excluding pensions). The trajectories adopted in the National Assembly and promulgated differ for 2028-2030 (68.3/72.8/76.3 versus 69.5/72.6/75.7): the cost schedule uses the promulgated law. Part of the military aid to Ukraine is funded in the programming law: overlap declared with ending bilateral aid. Initial trajectory of Law no. 2023-703 (article 4), before its update: €53.7bn / €56.9bn / €60.4bn / €63.9bn / €67.4bn from 2026 to 2030 (€400bn over 2024-2030); the update adds about €36bn over 2026-2030 (338.2 versus 302.3) and €8.3bn in 2030 (75.7 versus 67.4). Refusing these €36bn amounts to returning to the initial trajectory: €8.3bn in 2030, i.e. about 0.24 points of 2030 GDP (€3,467bn: 2025 GDP × 1.03^5, the same assumption as for 2032 GDP), maintained in 2032 (editors' calculation).

NATOcostable, spending
  • Leaving NATO: end of the contribution to the common budgets (direct gain only): a gain of €0.83bn per year (0.2 to 1) (coefficient calculated by the editors from Cour des comptes, « La participation de la France à l'OTAN : une contribution croissante » (S2023-0770)).
Current law, horizon, limitations and 2 sources

Current law. France is a member of the Alliance and has been back in the integrated military command since 2009. Contribution to NATO common budgets: €203m in 2022 (of which €193m from programme 178). Cost share of 10.1023% for 2026-2027 (10.1940% from 7 March 2024 to 31 December 2025). According to the Cour des comptes, the contribution could reach about €830m in 2030 in 2022 euros (€770m for the military and investment budgets, €63m for the civil budget). The common budgets of all Allies amount to about €4.6bn in 2025.

Horizon. Full year 2030-2032. The Cour des comptes costs in 2022 euros; the top of the range is uprated at about 2% inflation per year.

Limitations. Only the direct contribution is costed. Leaving NATO would not reduce the national defence target and could increase it (loss of pooling, of collective protection, of intelligence): not to be shown as a net saving. Low: 2022 level (0.2). Central: the Cour's estimate for 2030 (0.83, 2022 euros). High: 0.83 uprated for inflation (~1.0), editors' calculation. The Cour stresses the opacity of budget documents on this contribution. Leaving the integrated command alone has no source for a costing: not costed. Leaving the integrated command alone: no recent official source costs its budgetary effect.

  • Cour des comptes, « La participation de la France à l'OTAN : une contribution croissante » (S2023-0770), 2023-07: « « Pour la France, le financement des budgets militaires et d'investissement par le programme budgétaire 178 pourrait atteindre 770 M€ en 2030 et celui du budget civil par le programme budgétaire 105 un montant de 63 M€ en 2030. Ces chiffres exprimés en euros 2022 ne prennent pas en compte l'inflation à venir. » ; « la contribution financée par le programme 178 pourrait passer de 193 M€ en 2022 à environ 770 M€ en 2030 » »
  • OTAN, « Funding NATO » (page et tableau des quotes-parts), 2026-04-08: « These funds (around EUR 4.6 billion for 2025, and up to EUR 5.3 billion for 2026) enable NATO to deliver capabilities […] [tableau des quotes-parts, Cost share « at 32 », valid from 1 January 2026 until 31 December 2027] France 10.1023 »
Support for Ukrainecostable, spending
  • Ending bilateral aid: civilian share as central value, military aid added as upper bound (gain only if the appropriations are not redeployed): a gain of €0.24bn per year (0 to 2.3) (coefficient calculated by the editors from Ministère des Armées, Rapport au Parlement 2025 sur les exportations d'armement de la France (page 11 du PDF)).
  • End of bilateral civil aid only (humanitarian aid and Ukraine Fund), military aid maintained: a gain of €0.24bn per year (coefficient calculated by the editors from Ministère des Armées, Rapport au Parlement 2025 sur les exportations d'armement de la France (page 11 du PDF)).
  • French share of the 2028-2034 Ukraine facility proposed by the Commission (€100bn over 7 years), if it is financed by national contributions: a cost of €2.5bn per year (2.4 to 2.6) (coefficient calculated by the editors from Ministère des Armées, Rapport au Parlement 2025 sur les exportations d'armement de la France (page 11 du PDF)).
Current law, horizon, limitations and 8 sources

Current law. France's military support estimated at €5.9bn from 24 February 2022 to 31 December 2024 (including the €0.4bn bilateral fund), plus the French contribution to the European Peace Facility, up to €2.3bn of which €1.4bn suspended (2025 report to Parliament on arms exports); €6.5bn up to 31 December 2025, EPF included (up to €2.1bn), according to the 2026 report, which changes scope. The total of €8.6bn cited by an amendment (5.9 + 0.4 + 2.3) counts the bilateral fund twice. Annual pace according to the Minister for the Armed Forces: €1.7bn (2022), €2.1bn (2023), « au-dessus de 2 milliards mais en deçà de 3 » [“above 2 billion but below 3”] in 2024. Civilian aid: more than €434m in humanitarian aid at 1 September 2024 (Ministry for Europe and Foreign Affairs), Ukraine Fund of €200m (payment appropriations from 2024 to 2027) then €71m in the finance act for 2026. Franco-Ukrainian bilateral security agreement. EU: €90bn loan for 2026-2027 financed by European borrowing. The interest charge is about €3bn a year for the EU budget from 2028, shared among 24 States, Hungary, Slovakia and Czechia being exempt. For 2028-2034, the Commission proposes a €100bn Ukraine facility outside the ceilings.

Horizon. Full year, current €bn. 2022-2024 values for bilateral aid. The reference law in 2032 depends on the outcome of the conflict.

Limitations. A large part of bilateral military aid takes the form of transfers from stocks, funded in the military programming law. Stopping it frees up appropriations only if the defence budget falls by the same amount; otherwise, the budgetary gain is zero. The range for bilateral aid is calculated by the editors: low = 0 (« une grande partie de l'aide à l'Ukraine repose sur les équipements anciens que nous retirons du format des armées grâce à la LPM » [“a large part of the aid to Ukraine relies on old equipment that we are withdrawing from the armed forces' format thanks to the LPM”], according to the minister: without a cut in the defence budget, nothing is saved); central = the civilian share alone, about €0.24bn per year (€434m of humanitarian aid over 2.5 years, i.e. 0.17, plus the Ukraine Fund, €271m over 2024-2027, i.e. 0.07; a combination of two partial sources, as no official source gives a total for civilian aid); high = 2.3, the civilian share plus military aid excluding the FEP (European Peace Facility), €5.9bn over 2.85 years (€2.07bn per year), if military appropriations were also removed from the budget. The former central value (0.95, difference between €8.6bn and €5.9bn) had no basis: this difference is not civilian aid. The former upper bound (€3bn) took up a statement (« plus près de trois » [“closer to three”]) that the official record corrects (« en deçà de 3 » [“below 3”]), and which included about €0.3bn of revenue from frozen Russian assets, paid by the EU. The French share of the interest is a calculation: €3bn × 17.2% to 18% (2024 key, raised because three States are not participating). Commissioner Serafin puts this interest at about €1bn in 2027 and €3bn per year from 2028. The 2028-2034 facility is only a proposal: 100/7 ≈ €14.3bn per year × ~17%. No official assessment of 2025 aid was found. In 2032, the reference law depends on the conflict; cost schedule convention: latest observed level maintained. The French share of the interest on the €90bn European loan is already committed (current law). Option arret-aide-civile-bilaterale: the civilian share alone (€0.24bn per year, same calculation as the central value for ending bilateral aid), for a position that maintains military aid.

Overseas territories

High cost of living overseascostable, spending
  • Abolition of the octroi de mer (dock dues) with full compensation by the State to local authorities: a cost of €1.6bn per year (1.55 to 1.85) (coefficient calculated by the editors from Banque des Territoires, « L'octroi de mer ne contribue que marginalement à la vie chère, selon un rapport » (étude AMF / ACCD'OM)).
  • Regulatory measures (caps on margins, quality-price shield, transparency): no direct budgetary cost: a cost of €0bn per year.
  • Abolition of the octroi de mer (dock dues), compensated or not (revenue loss for general government as a whole): a cost of €1.6bn per year (1.55 to 1.85) (coefficient calculated by the editors from Banque des Territoires, « L'octroi de mer ne contribue que marginalement à la vie chère, selon un rapport » (étude AMF / ACCD'OM)).
Current law, horizon, limitations and 4 sources

Current law. The octroi de mer, a tax on imports and local production levied by the authorities of the overseas departments and regions (DROM), raised €1.545bn in 2023 and €1.644bn in 2022 (record). It accounts on average for 32% of the operating revenue of DROM municipalities. Its regime expires at the end of 2027 (European decision) and must be reformed. The bill to combat the high cost of living in the overseas territories (presented on 30 July 2025, adopted by the Senate on 28 October 2025, AN no. 2046) has had its examination postponed and has not been promulgated as of the date of the sources consulted.

Horizon. Full year 2032, current €bn. High = 2023 yield uprated by 2% per year until 2032.

Limitations. The effect of abolition on prices is debated: in the illustration in the elected officials' study (AMF), the octroi de mer (dock dues) paid by the importer represents 4.4% of the final price and that paid by the consumer 7.9%; the yield of €1.545bn in 2023 is taken from the press (Banque des Territoires, based on OFGL data), not from that study; an upward effect « plus dur pour les ménages modestes » [“harder on low-income households”] according to the Cour. Compensation by the State is the most costly assumption; replacement by another local tax (regional VAT) would cost the State ~0. The 2024 yield was not found. The range is calculated by the editors: low = 2023, central = 2022-2023 average, high = 2023 × 1.02^9. The fate of the bill after the postponement of February 2026 is not confirmed (legislative file not consulted). The balance is that of general government as a whole: abolishing the octroi de mer costs its yield there, whether or not the State compensates local authorities (compensation is an internal transfer). The “suppression-octroi” option therefore takes the amounts of the compensated option, and is used when the text says nothing about compensation; the two options are mutually exclusive (overlap declared). Same range and same sources: €1.545bn in 2023 (Banque des Territoires), more than €1.6bn in 2022 (Maire-info, Cour des comptes).

Status and autonomy of the territoriesnot costable
Current law, horizon, limitations and 2 sources

Current law. Constitution: the DROM are governed by art. 73 (legislative identity, adaptations possible), the overseas collectivities by art. 74 (own status set by organic law) and New Caledonia by Title XIII. New Caledonia: the Bougival agreement (July 2025), supplemented by the Élysée-Oudinot agreement (January 2026), provided for an « État de la Nouvelle-Calédonie » [“State of New Caledonia”] and a Caledonian nationality. The constitutional bill, adopted by the Senate on 24 February 2026, was rejected by the National Assembly on 2 April 2026 (preliminary motion to reject, 190 votes to 107).

Horizon. Not applicable.

Limitations. No identifiable static budgetary effect. Transfers of powers come with financial transfers that depend on negotiation. The follow-up after the rejection of April 2026 is not verified at the date of the cost schedule.

Society

Secularism (laïcité)not costable
Current law, horizon, limitations and 1 source

Current law. Constitution, art. 1 (« République indivisible, laïque, démocratique et sociale » [“indivisible, secular, democratic and social Republic”]); act of 9 December 1905 (separation of the Churches and the State); act of 15 March 2004 (religious symbols in state schools); act no. 2021-1109 of 24 August 2021 reinforcing respect for the principles of the Republic. Alsace-Moselle: Concordat regime maintained.

Horizon. Not applicable.

Limitations. Lever of principles, with no budgetary effect. Sole exception: ending the Alsace-Moselle concordat would abolish the State's payment of ministers of religion, but no figure was found in this research. The references to the laws of 1905, 2004 and 2021 are cited without online checking.

  • Constitution, art. 1er: « « La France est une République indivisible, laïque, démocratique et sociale. » »