France’s 2027 budget: household losses and consumption risk

In the simulation published by the OFCE on 16 October, households at the bottom of the living-standard distribution lose about €100 on average in 2027, compared with about €500 at the top. Measured against their resources, the ranking reverses: the loss amounts to roughly 0.8% of the first group’s living standard, against 0.35% for the last. These two views of the same budget proposals describe very different experiences of the burden. The economic consequences then depend on adjustments between everyday purchases, saving and postponed expenditure. ¹

Pierre Madec’s analysis estimates a €8.7 billion reduction in household disposable income in 2027 for the measures it examines, relative to its reference scenario. It provides a useful map of their initial incidence. Assessing the associated risk means following the money through consumption, businesses and public revenue, while also considering how a fragile debt trajectory could raise financing costs. ^{1, 12, 13, 15}

The investigation starts, however, with a timing problem. The simulation covers measures from the government’s original proposals, some of which the National Assembly had already removed before the study appeared. The figures below retain their status as results for that scenario. Parliamentary developments are identified separately, using the documents available on 20 October 2026. ^{1, 5, 6, 7}

A figure tied to a particular version of the budget

France’s finance bill and social security financing bill were introduced on 1 October. The finance bill included a €3,000 sub-ceiling on the portion of the 10% income-tax allowance attributable to retirement pensions. On 15 October, MPs adopted amendments deleting Article 3, which contained that proposal. The deletion remains in the Assembly’s provisional consolidated document issued on 20 October, incorporating proceedings through the second sitting of 19 October. The document is provisional, and the legislative process continues. ^{2, 3, 5, 6}

On 14 October, amendment I-502 and identical amendments had also removed provisions increasing income tax and the CSG social contribution on certain sickness and occupational accident or disease daily benefits. The OFCE discusses these in a supplementary section, outside its main charts. Any account of them therefore needs to identify their place in the original proposal. The main €8.7 billion figure already excludes these daily benefits; it does include the pension tax-allowance ceiling that MPs removed on 15 October. ^{1, 6, 7}

That chronology matters for the calculation. The OFCE attributes about €1.1 billion of lost disposable income to the pension allowance proposal. Subtracting it from the headline figure and announcing an updated “cost of the budget” would imply more precision than the evidence supports. A new estimate would require another simulation using the relevant text, the interactions between taxes and benefits, and the other changes to the package. The published result describes one configuration; a different configuration requires a fresh calculation. ^{1, 9}

The scope also needs to remain visible. The study examines several measures affecting households directly, principally through pensions, public-sector pay and benefits. Other budget decisions, including those transmitted through businesses, public services and investment, belong in a broader assessment. This investigation uses the 2027 budget to examine a specific transmission mechanism within a defined set of measures. ^{1, 2, 3}

A freeze opens a gap between two paths

Understanding a freeze requires comparing two trajectories. Along one, a benefit rises according to the uprating assumed in the counterfactual, the reference path chosen in the absence of the freeze. Along the other, its amount stays unchanged. The shortfall is the difference between those paths, even when the payment reaching a household’s account remains identical to the previous year’s. Nominal income growth, purchasing power and the difference from a reference policy are therefore separate measurements. ¹

The OFCE assumes, among other things, a 2.3% April increase for the prime d’activité, France’s in-work benefit, and for the monthly family-benefit base (BMAF). It also assumes a 2% October increase in the expenditure parameters used to calculate housing benefits. That last percentage is an assumption made for the study. Article 74 of the finance bill proposes suspending the uprating of the in-work benefit and housing-benefit expenditure parameters, while allowing their restoration by decree. Housing-benefit income parameters reviewed in January follow a separate timetable. For family benefits, the government’s press dossier separately proposes adjusting their calculation base on 1 April to neutralise statutory uprating. ^{1, 2, 4}

The starting month makes a large difference to the first-year amount. A gap beginning in April operates for nine months of 2027; one beginning in October operates for three. The OFCE estimates the in-work benefit freeze at €723 million in 2027 and €965 million over a full year. The corresponding housing-benefit figures are €81 million and €323 million. These amounts describe the same mechanism over different periods. The full-year measure is a twelve-month comparison under the study’s assumptions; treating it as an independent forecast for 2028 would overlook subsequent policy decisions and economic changes. ¹

Public-sector pay introduces a more fundamental distinction. The study compares a frozen salary index point with a hypothetical 1.8% rise, matching its assumed inflation rate for 2027. That pay component has no automatic statutory indexation. The roughly €1.7 billion income shortfall estimated by the OFCE is consequently a gap relative to this assumed increase, a distinction the author explicitly makes. Its interpretation depends on keeping both the chosen pay baseline and the coverage visible: the study includes the three branches of the French public service and measures households’ net income. ¹

The €8.7 billion total thus combines gaps with different institutional origins. It measures their income consequences under the specified scenario. Translating it into an improvement in the public balance would require reconciling gross spending, taxes, social contributions, transfers between public bodies and reference assumptions across a consistent scope. That reconciliation comes before any fiscal-multiplier calculation. ^{1, 9, 12}

Cash amounts and the burden on household resources

The French statistical concept of niveau de vie adjusts income for household size. Insee defines it as disposable income divided by consumption units: one for the first adult, half a unit for each additional person aged 14 or over, and 0.3 for a younger child. This recognises that some costs, including housing, can be shared. Households with identical total incomes can therefore have different living standards. In English, the measure is generally called equivalised disposable income. ⁸

The OFCE groups households into twenty ordered bands of living standards, or ventiles. Average losses in euros tend to rise towards the top of the distribution. Their relative burden follows a different profile: around 0.8–0.9% in several lower groups, falling to 0.35% at the top. The more affluent half can bear about 64% of the total cash loss while lower groups experience a heavier proportional burden. Those findings describe different dimensions of the same scenario. ¹

Averages also conceal how widely losses are spread. The study identifies 22.4 million losing households, with an average loss of about €390 among them. The ventile averages in the first chart, by contrast, cover all households in each group. Confusing the two would understate the loss for those actually affected. In the published tooltips, more than 85% of households lose in groups three to eight, compared with about 59% in the top group. ¹

Several measures can affect the same household. A lower-paid household may receive the in-work benefit, family benefits and housing assistance. Protecting the uprating of a minimum-income payment therefore leaves a separate question about its other resources. Similarly, protection for a small pension is defined at the beneficiary level, whereas living standards depend on all household income and household composition. Distributional analysis needs to follow those different units of assessment through the calculation. ^{1, 3, 8}

Different instruments reach different households

A simple division between pensioners and working-age households is too broad to explain the package. In the simulation, the more affluent half bears 96% of the loss from the pension tax-allowance ceiling. For the back-to-school allowance reduction modelled by the OFCE, the direction reverses: the less affluent half bears 96% of the loss. Its share reaches 87% for the in-work benefit freeze and 95% for housing benefits. The profile of the affected households tells us more about a measure’s incidence than its administrative label. ¹

Removing the pension allowance ceiling therefore removes, at this stage, an instrument whose burden was concentrated near the top of the distribution. That establishes the direction of this particular change, holding other things constant. The overall distribution will depend on the rest of the legislation and any replacement measures. The available evidence identifies the instrument that was removed; the complete picture of the final budget remains to be calculated. ^{1, 5, 6}

The back-to-school allowance raises a different issue about the scenario’s construction. The government records an intended saving and proposes retargeting the allowance, with parameters to be determined after consultation and attention to children’s ages. To make the measure calculable, the OFCE assumes a uniform 30% reduction, generating a €550 million disposable-income loss within its scope. The 30% cut is therefore a modelling assumption. Targeting by age or changing eligibility thresholds could affect different families, even if the total saving were similar. ^{1, 4}

For pensions, Article 35 of the social security financing bill is also more detailed than a single cliff edge at €1,260 a month. It provides for full uprating up to that amount of total pension income, a transition up to €1,281, and the possibility of coefficients set by decree above the threshold, with additional income bands. The derogation concerns basic-scheme pensions; total pension income determines the beneficiary’s position. The assumed freeze and treatment of supplementary pensions must therefore be distinguished from the options the draft legislation allows. ³

Where households absorb the shortfall

A reduction in disposable income can lead to lower purchases, reduced saving or the use of money previously set aside. The split depends on the resources that are actually available when the shock arrives, its expected duration and the expenses that are difficult to postpone. A home is an asset, but its value cannot immediately become cash to meet the month’s bills. This distinction between wealth and liquidity is one reason the distribution of the burden matters for the economy. ^{14, 16}

Economists use the marginal propensity to consume to describe the share of a change in resources reflected in consumption over a specified period. A 2025 European Central Bank (ECB) study estimates different responses across income groups in an analysis of mortgage transmission. Its design, population and type of shock limit its transferability to frozen French benefits. The response to the 2027 measures still needs to be estimated: taking an international coefficient and multiplying it by €8.7 billion would create the appearance of precision without establishing the connection between the two. ¹¹

Pensioner status alone supplies too little information. One household may have a regular income and very limited liquid reserves; another may temporarily absorb the shortfall by reducing contributions to a savings product. Research by Austria’s central bank (OeNB) on heterogeneous responses to additional income reinforces the need to examine household balance sheets. It uses a hypothetical lottery-windfall question and finds differences across countries, so its conclusions must remain anchored in that survey design. ¹⁶

The national saving rate cannot resolve the question either. Banque de France research shows that the composition of income, particularly financial income, helped sustain high saving in 2023–24. An aggregate average can therefore rise while households affected by a particular measure have little room to adjust. For the budget assessment, the relevant information would be the capacity of those specific households to absorb the shortfall when it occurs. ¹⁰

From household finances back to the public balance

The next stage depends on which purchases households actually reduce. Part of their value pays for French production; another part reflects imports. Insee’s work on the content of household consumption shows large differences across products. That historical picture helps explain the mechanism, while a 2027 estimate would require identifying the spending cut at the margin. A product’s origin also leaves a separate question about public revenue: its sale can generate French distribution margins and VAT. ¹⁷

Businesses facing weaker demand may then adjust output, investment or employment. The resulting income changes feed into further consumption decisions. Comparisons of the French Mésange and ThreeME models published in July 2026 illustrate these links and their dependence on the policy instrument and time horizon. A transfer to a household works first through its spending decision, whereas a public order creates demand directly. Applying the same coefficient to both would erase that distinction. ¹²

The process then feeds back into public finances. Weaker tax bases can reduce receipts, while some spending, including unemployment benefits, can rise. This is one role of the automatic stabilisers described by the ECB. Their operation explains why the yield after the economy responds can be smaller than a measure’s initial yield. Quantifying that difference requires an estimate for the particular scenario, including its tax structure, timing and other economic responses. ¹³

This feedback also creates a methodological problem when assessing a budget forecast. An official projection may already incorporate some effects of its measures on consumption, employment and revenue. Adding a separately calculated macroeconomic loss could count the same effects twice. The useful comparison is between two internally consistent paths, with and without the package, under the same external assumptions. The assessment then needs to specify what each path includes and which responses remain outside the model. ^{9, 12, 13}

The OFCE microsimulation stops before these reactions. Ines documentation describes its standard use as a static model of immediate monetary effects; behavioural responses and macroeconomic feedback require further work. The €8.7 billion figure consequently informs the initial incidence on income. Corresponding estimates of lost consumption, lost GDP and foregone public revenue remain to be produced. ^{1, 9}

Financing costs are another part of the risk

Our analysis of bond prices and sovereign refinancing explains the passage from market yields to government interest payments and includes a sensitivity tool.

The counterargument follows a different route. A fiscal trajectory judged less sustainable can raise the return lenders demand. As debt is refinanced, a larger share of public resources then goes towards interest. France’s High Council of Public Finances, the HCFP, reports a government sensitivity calculation: a permanent 100-basis-point increase, equivalent to one percentage point, across all maturities from 1 January 2027 would add €3.4 billion to interest costs in the first year and €8.1 billion in the second. These are annual additional costs under the assumed rate shock. ¹⁵

That sensitivity shows why financing belongs in the analysis. Estimating the rate benefit attributable to a particular benefit freeze or tax-allowance change would require a separate study. Between the choice of a measure and the cost of public debt lie the overall fiscal path, its credibility, inflation, monetary policy and market conditions. This transmission has its own timetable, just as the transmission from income to consumption does. ^{14, 15}

An International Monetary Fund (IMF) working paper published in June 2026 examines this trade-off in a model of a highly indebted economy calibrated using Belgium, France and Italy. Its results depend on the assumed path without adjustment and on the financing response. Crucially, its expenditure-led package increases targeted assistance to vulnerable households while reducing other transfers. That composition offers another reason to examine who bears the burden. The exercise provides a theoretical counterpoint under specific assumptions, rather than a forecast of France’s proposals. ¹⁴

The HCFP locates this tension within the government’s 2027 submission. It considers the 1% growth forecast optimistic, with stronger private demand expected despite a larger fiscal adjustment. The credibility of the fiscal effort and the credibility of the activity forecast therefore need to be examined together. An announced saving is more convincing when its implementation, distribution and assumed behavioural response can all be explained. ¹⁵

Reassessing the package after parliamentary votes

The useful next step would be to reconstruct the package retained after parliamentary consideration, including the decrees and parameters needed for implementation. Deleted measures, restored upratings and the final back-to-school allowance rules would change the starting calculation. The analysis would then need to reconcile household disposable income with spending and receipts across public bodies before connecting those results to a path for consumption and activity. ^{1, 2, 3, 4, 6, 9}

That assessment should document the liquid resources of affected households, how long they expect the shortfall to last and which expenses they adjust. It could then compare different distributions of the same adjustment, tracing their consequences for households, activity, financing and the public balance. The models examined here show why those dimensions can interact; applying them to the 2027 package remains a separate task. ^{9, 12, 14}

The distribution of the burden belongs inside the economic assessment of the budget. It determines the resources available where spending and saving decisions are made. The OFCE’s data make that initial incidence visible, and parliamentary votes have already changed some of its components. Assessing the eventual saving requires following those changes through household responses and financing costs, while keeping the reference scenario consistent at every stage. ^{1, 5, 6, 12, 14}

Method and interpretation

The charts reproduce rounded values from the OFCE page’s tooltips, extracted on 20 October 2026. The publication contains some internal discrepancies, including €500 for the top ventile in its tooltip against €510 in the fixed annotation, and 0.83% for the first ventile against 0.84% in the prose. We retain the extracted series without silently adjusting it. Individual records, weights and study-specific programs are not supplied with the post. The extraction reproduces displayed results and supports arithmetic checks; it does not reproduce the microsimulation itself. ¹

The study uses Ines, drawing on the 2022 Revenus fiscaux et sociaux survey and a 2024 legislative base. Updating incomes and weights forms part of Ines’ general methodology, so the survey year should be read alongside the simulation protocol. Its usual scope covers households in private accommodation in metropolitan France. The references below identify the passages used and their scope. No interviews or exchanges with the authors or public authorities were conducted. ^{1, 9, 18}

The opening illustration is a conceptual image generated with AI.

Sources and documents consulted

1. S01 · Pierre Madec · OFCE · 2026-10-16 · Accessed 20 October 2026.Budget 2027: an unevenly distributed household burdenFull post, Table 1, Charts 1–3, methodology box and notes. Static microsimulation of the original package; charts exclude daily sickness benefits. Rounded displayed data, with internal discrepancies explained in the Method section. Charts redrawn by l0g from the published series under CC BY 4.0.
2. S02 · Assemblée nationale · 2026-10-01 · Accessed 20 October 2026.2027 finance bill, No. 3210 : original textArticles 2, 3 and 74, including explanatory statements. Introduced version. Article 74 covers the in-work benefit and housing-benefit expenditure parameters, with possible restoration by decree. The family-benefit base is a separate proposal.
3. S03 · Assemblée nationale · 2026-10-01 · Accessed 20 October 2026.2027 social security financing bill, No. 3211Article 35 and its explanatory statement; fiscal trajectory annex. Basic pensions: full uprating up to total monthly pension income of €1,260, transition to €1,281 and possible decrees above the threshold. Original proposal, distinct from enacted law.
4. S04 · Gouvernement · Sécurité sociale · 2026-10-01 · Accessed 20 October 2026.Budget 2027 : government press dossierSocial security section, printed p. 22 (PDF page 144), family benefits and back-to-school allowance; PDF page 30, public pay. Proposed neutralisation of family-benefit uprating through an adjusted base; back-to-school allowance parameters subject to consultation. The uniform 30% reduction is an OFCE assumption.
5. S05 · Assemblée nationale · 2026-10-15 · Accessed 20 October 2026.Public vote No. 8738, deletion of Article 3First sitting of 15 October, vote subject and adoption of deletion amendments. Establishes deletion of the pension allowance sub-ceiling in the Assembly sitting. The legislative process continues.
6. S06 · Assemblée nationale · Direction de la séance · 2026-10-20 · Accessed 20 October 2026.Provisional consolidated 2027 finance bill as of 20 OctoberCover: proceedings through the second sitting of 19 October. Article 2, pp. 2–7; Article 3 deleted, p. 8. Explicitly provisional. The file may be replaced at the same URL; the dated copy and hash identify the version examined. No claim is made to cover every vote on 20 October.
7. S07 · Assemblée nationale · Estelle Mercier and co-signatories · 2026-10-14, adoption; tabled 8 October · Accessed 20 October 2026.Amendment I-502 to Article 2 of the 2027 finance billAdopted status on 14 October; operative text and explanation. Second sitting record of 14 October for the vote with identical amendments. Deletes the increased income tax and social levies on the relevant daily benefits. Introduced on 8 October, adopted on 14 October.
8. S08 · Insee · 2021-10-18 · Accessed 20 October 2026.Definition: equivalised disposable incomeDefinition and equivalence scale. Household disposable income divided by consumption units. Distinguishes total income, living standards and household composition.
9. S09 · Insee · Drees · Cnaf · Publication date not stated · Accessed 20 October 2026.Ines model documentationIntroduction, coverage and static nature of the model. Immediate monetary effects in standard use, excluding behaviour and macroeconomic feedback. Additional work and modules may address other questions.
10. S10 · Alice Carroy, Camille Thubin · Banque de France · 2025-03-21 · Accessed 20 October 2026.How far has financial income supported France’s household saving rate?Blog post 397, income composition and saving in 2023–24. Income composition and saving in 2023–24; a different scope from responses to the 2027 policy package.
11. S11 · BCE · Economic Bulletin 4/2025 · 2025 · Accessed 20 October 2026.The transmission of monetary policy: from mortgage rates to consumptionSection 4 and footnote 10, consumption responses and MPC construction. Mortgage cash-flow channel and transitory shocks. No coefficient is applied to the French package.
12. S12 · DG Trésor · OFCE · 2026-07-09 · Accessed 20 October 2026.Comparison of the properties of the French ThreeME and Mésange modelsSections 1.1.1 and 1.4.7, then 2.2–2.3: constant real rate assumption, public finances, expenditure and VAT experiments. Policy experiments by instrument and horizon, distinct from a simulation of the 2027 bill. They do not quantify sovereign credibility effects. Attached PDF consulted.
13. S13 · BCE · Economic Bulletin 6/2020 · 2020 · Accessed 20 October 2026.Automatic fiscal stabilisers in the euro area and the COVID-19 crisisSections 2–4: tax bases, contributions, unemployment and stabilisation. Used for general mechanisms. The simulations in the 2020 setting are not applied to the 2027 budget.
14. S14 · Hong, Mohiuddin, Ouedraogo, Smirnov, Vaziri · FMI WP/26/121 · 2026-06 · Accessed 20 October 2026.HANK-Based Fiscal Consolidation for a High-Debt Advanced Euro Area EconomyIntroduction, footnote 1 on printed p. 6, calibration and consolidation scenarios. Representative Belgium–France–Italy economy. The expenditure package increases targeted assistance. Results depend on debt and sovereign-premium assumptions; an authors’ research paper.
15. S15 · Haut Conseil des finances publiques · 2026-09-25 · Accessed 20 October 2026.Opinion HCFP-2026-5 on the 2027 finance and social security billsParagraph 44, p. 14, growth forecast; paragraphs 115–116, pp. 38–39, interest costs and rate sensitivity; Chart 20. Opinion released on 1 October. Government permanent +100 bp all-maturity shock from 1 January 2027: +€3.4bn in 2027 and +€8.1bn in 2028, annual additional costs. No rate benefit is attributed to the measures examined.
16. S16 · Albacete, Fessler, Pekanov · OeNB Working Paper 254 · 2024 · Accessed 20 October 2026.The Role of MPC Heterogeneity for Fiscal and Monetary Policy in the Euro AreaHFCS methodology, Table 2 and French Appendix A4.6. Reported responses to a hypothetical lottery windfall. Cross-country and specification differences preclude a uniform MPC for French pensioners. No coefficient reused.
17. S17 · Alexandre Bourgeois, Antonin Briand · Insee Première 1756 · 2019-06-05 · Accessed 20 October 2026.French content accounts for 81% of total household consumption, but only 36% of manufactured goods2015 data; imported content by product and consumption definitions. Historical snapshot used to explain differences in content. No aggregate share is applied to marginal spending in 2027.
18. S18 · Insee · Drees · Cnaf · 2025-02-03 · Accessed 20 October 2026.Introducing the Ines modelStatistical inputs, updating, simulated benefits and taxes. Documents the general method. Generic code availability is insufficient to reproduce the specific OFCE exercise without its data and parameters.