The same budgetary effort of €10 billion, allocated using INSEE’s 2023 distributional accounts (an accounting profile at constant structure, not a simulation of its effects), falls on different households depending on the instrument chosen: an adjustment through education weighs proportionally more on low-income households, an adjustment through taxes on income and wealth more on well-off households. Who pays for public debt depends on what it finances, how it is financed and the adjustments chosen to service it.
The question arises in every indebted country. France is a useful case study for it: its statistical office, INSEE, publishes distributional national accounts that allocate taxes, contributions and public transfers across the income scale, which makes it possible to ask not only how much the debt costs, but how a given adjustment would be distributed across households.
Who pays for public debt? The answer changes with the decision taken to adjust it. In an accounting profile drawn from INSEE’s accounts, the same €10 billion adjustment weighs mostly on the least well-off households if it passes through education, mostly on the best-off if it passes through taxes, and far less unevenly across deciles if it passes through pensions. The size of the bill — stock, interest burden, average cost of the stock — is measured on another page of this dossier, What does French public debt actually cost?; this one asks on whom it falls, what the data do, or do not, allow us to attribute, and what a real decision, considered for 2026 and then set aside, shows.
How is the same adjustment shared out depending on the lever chosen?
The population is ranked from the least to the most well-off, then cut into ten groups of equal size, the deciles of usual standard of living (disposable money income per consumption unit). The exercise distributes €10 billion three times: through a rise in taxes on income and wealth, a cut in retirement pensions, and a cut in education spending. Each time, it allocates the sum in proportion to what each decile already pays or receives, then relates the result to that decile’s income. The tax considered is the distributional accounts’ category of “taxes on income and wealth”: the CSG (a broad-based social contribution on income), income tax and the other taxes in this category, such as the CRDS, the housing tax or the tax on real-estate wealth. It is neither a forecast nor a measure under consideration: no behaviour and no feedback effect enter it. It is the structure of these items in 2023, the latest year published, made comparable from one decision to another.
INSEE, distributional national accounts, table CND.101 (standard-of-living twentiles, 2020-2023, base 2020); 2023 vintage. Accounting profile at constant structure, without modelling behavioural responses or feedback effects.
The result in one sentence. “Who pays for the debt?” has no answer in itself; it has one per decision: in an accounting profile drawn from INSEE’s 2023 accounts, the same adjustment weighs mostly on the least well-off if it passes through education, mostly on the best-off if it passes through taxes.
The ranking reverses from one decision to another. In this accounting profile, a cut in education spending would represent 2.75% of the income of the poorest 10% and 0.21% of that of the richest; a rise in taxes on income and wealth would do the opposite, up to 1.14% for the top decile. The contrast does not come only from the lower deciles’ smaller incomes: in euros, the less well-off half would bear 57.8% of the cut in education spending, against 14.6% of the tax increase. A cut in pensions would stay within a narrow band, from 0.43 to 0.78%, sparing no group in particular. And “taxes” do not have a single profile: a rise in taxes on products, such as VAT, distributed in the same way, would also weigh more on the least well-off as a share of their income, 1.14% against 0.43% for the best-off. The profile favourable to the least well-off is therefore that of taxes on income and wealth, not of all taxation; and this calculation follows the same proportional convention, without being the final incidence (who ultimately bears the burden, not who remits it) of an actual VAT rise.
An accounting exposure, not an observed loss. The 2.75% for the first decile is the share of the effort that falls to it, allocated in proportion to the education services that INSEE’s accounts attribute to it by convention, relative to its income: neither a loss of income actually suffered, nor a measure of well-being. A tax rise takes away money; a spending cut takes away a service, whose value is imputed and whose real cost for each person depends on what is cut.
In the middle of the scale, the choice of lever matters least. For the 7th or 8th decile, a little above the middle, the average burden would change little from one decision to another: the heaviest would cost 1.4 times the lightest there. For the richest 10%, it would cost 5.5 times the lightest, and 5.5 times for the poorest 10%: at both ends of the scale, the choice of lever decides almost everything. These deciles would not bear less of the adjustment; changing the lever would simply alter their average exposure less, which does not describe the situation of each household in them. The convergence appears in all four vintages 2020-2023, published by INSEE as experimental statistics (which share conventions and structures, and are therefore not independent replications), and it holds only for these three levers distributed in this way.
The trough in D7-D8 put to the test, and the limits of the exercise
In twentiles. With the population split into twenty groups instead of ten, the minimum still falls in the D7-D8 zone: in the 14th or 15th twentile depending on the year. The trough does not come from grouping by deciles.
By pairs of levers. The tax curve rises with the standard of living, the education curve falls: taken two at a time, they come closest in D7, and it is their crossing that makes the trough; the pensions curve, almost flat, passes above both at that point. The coefficient of variation and the range relative to the mean place the minimum in the same zone as the ratio of the heaviest to the lightest.
Measured in percentage points of income rather than as a ratio, the gap between the heaviest and the lightest decision is still smallest in D7-D8. Without pensions, taxes and education alone come closest in D7: the trough owes nothing to the third curve. With a fourth lever, taxes on products, it stays in D7-D8.
With other forms of measures (calculation assumptions, not measures under consideration). An education cut of the same amount for each decile, or concentrated where spending is already high, leaves the trough in D7-D8; sparing the education of the first three deciles places it in D8 or D9 depending on the year; sparing their pensions leaves it in D7-D8. A more progressive tax, distributed in proportion to the square of what each decile already pays, moves it to D9, however. The trough therefore withstands the choice of measure and several distributions, but its location follows the progressivity of the measures: it is a result of the proportional scenario, not a general property of redistribution.
The reference income only affects the level of the curves. For a given decile, it is common to all three levers: the ratio of the heaviest to the lightest, and the points where two curves cross, do not depend on it. The gap measured in percentage points of income, however, does: with another denominator, its minimum could move.
The limits. A public service valued in euros is not income: education spending imputed to a household is not paid to it, and cutting it would not cost the household exactly that sum. The three decisions are proportional by construction; the figure says nothing about “taxes” or “public spending” in general. Finally, the first-decile point is excluded for the tax decision: income tax of the first twentile makes up 60% of the item there while the neighbouring twentiles are at zero, an atypical cell in the source. The gap for this decile, computed on the two other levers, is therefore a lower bound.
Who borrows, who holds, who receives the interest?
Four questions are often confused: who borrows, who finances and receives the interest, through which mechanisms financing and adjustments operate, and who ultimately bears their effects. These are not four populations: the same person pays taxes, is treated in a public hospital and holds government securities through an assurance-vie policy. And “who pays” is always said relative to another decision: borrowing less would have meant another tax, another cut in spending, with their own losers.
Two distinct scopes: A at nominal value (INSEE), B at market value (Banque de France, via Agence France Trésor). The figure on p. 101 of the book, closed on end-2025 accounts, gives an earlier version of it.
Who borrows, who holds. Central government accounts for most of the public debt: 2,942 billion euros out of 3,595.5, or 82% (end of June 2026). In Q2 2026, its marketable securities (medium- and long-term bonds and short-term Treasury bills, which trade on the market) were held 59.7% by non-residents; resident banks, insurers and funds together held 21.7%; the other resident holders, 18.7%, include the Banque de France, which the AFT chart used in the figure does not single out. The two panels cannot be combined: one is at nominal value (the amount the borrower will have to repay at maturity), the other at market value (the securities’ market price on the reporting date).
Who receives the interest: what is observed, what can be approximated, what is unknown. We observe the direct holders, classified by residence: a stock captured at a given date, not the flow of coupons, which depends on the securities held and when they were issued. We can approximate the indirect holders: tracing the investments of assurance-vie policies, funds and banks to their final use, the French Treasury (Direction générale du Trésor) estimates that out of every 10 euros of French households’ financial wealth, 1 finances general government debt, mostly French (Trésor-Éco no. 378, January 2026, in French). We do not know which households receive that interest: INSEE’s distributional accounts allocate it only by convention, in proportion to each group’s financial income, and no published statistic observes it, which would require following each security, its return and its intermediaries. Two readings are therefore ruled out: 59.7% of the securities is not 59.7% of the interest, and holdings of financial assets by decile do not tell each decile’s share of that interest. The absence of this breakdown does not mean that this income is negligible: it limits what can be asserted. Two further points: today’s holder is not necessarily the one who lent at issuance, since a security is then traded on the market; and the securities held by the Banque de France under Eurosystem programmes (the Eurosystem being the ECB and the national central banks of the euro area) cannot be analysed like those of a private creditor.
Residence is not nationality, and financing is not gaining
Financing is not gaining. The creditor advanced the funds: interest pays for that advance, for time and for risk. Its real return takes account of inflation and of changes in the security’s value, and its advantage over another investment is measured at a comparable horizon and risk. “We owe it to ourselves” is therefore only partly true, and does not say which of “us” receive and which finance.
The published breakdown classifies holders by residence, not nationality: a “non-resident” can be a foreign fund managing the savings of French households, and a French household can hold government securities without knowing it, through its assurance-vie policy. The Banque de France statistic on the long-term securities of general government as a whole covers a different scope from Agence France Trésor’s, which is limited to central government’s marketable securities: the two shares are not interchangeable.
Who contributes, who receives today?
INSEE’s distributional national accounts allocate, for 2023, taxes, contributions and public transfers across deciles of usual standard of living — the population ranked from the least to the most well-off, then cut into ten groups of equal size. Amounts are expressed per consumption unit (CU), the unit that makes households of different sizes comparable: a single person counts as 1 CU, a couple with two children under 14 as 2.1.
INSEE, distributional national accounts 2023 (Insee Analyses no. 118, 16 April 2026, figure 1c), in euros per consumption unit.
The values as a table (euros per consumption unit, 2023)
Decile
Taxes and contributions
Cash benefits
In-kind transfers
D1
−6,400
8,000
19,100
D2
−8,800
10,600
20,500
D3
−11,900
11,700
17,400
D4
−15,100
11,500
17,200
D5
−18,200
12,000
15,000
D6
−21,500
12,600
15,000
D7
−25,700
12,800
14,700
D8
−31,700
13,400
13,000
D9
−41,100
14,600
13,600
D10
−89,800
18,500
12,900
All
−27,000
12,600
15,800
Taxes and contributions range from €6,400 per CU for the poorest 10% to €89,800 for the richest 10%, about 14 times more. Transfers received — cash benefits and public services valued in euros — vary much less: they stay between €26,400 and €31,400, with no regular progression. What is paid tracks the standard of living; what is received, much less so. These taxes and contributions are not payments recorded one by one: besides income tax and social contributions, they include taxes on products, such as VAT, and taxes on production, which INSEE attributes to households by convention.
Where the debt enters these accounts. In 2023, transfers attributed to households exceed taxes and contributions by 66 billion euros, and INSEE links this gap to public borrowing: it is the part of the year’s extended income (the income of the whole economy attributed to households, after taxes, contributions and public transfers) that is not financed by the same year’s taxes and contributions. These 66 billion are not the 2025 interest burden (65.7 billion): year, scope and nature differ, and the closeness of the two figures is a coincidence.
A balance allocated by convention, financed on credit, which does not say who will pay tomorrow
This total covers the scope of the distributional accounts, which is not that of the public deficit: the two amounts are not interchangeable. In this construction, the gap raises households’ extended standard of living (that extended income per consumption unit) for the year (a present benefit, financed on credit), but these are not 66 billion of identifiable payments: it is a balance allocated by convention, which INSEE attributes half to lower taxes and contributions and half to additional transfers when it computes a standard of living “net of borrowing”. The figures on this page show current transfers, before that adjustment. The balance does not imply that the beneficiaries of redistribution “created” this debt, nor that their position says who will repay it.
INSEE, distributional national accounts 2023 (figure 2a). Two units, two panels: euros per consumption unit, then share of people.
Net contributors, net beneficiaries. On average per CU, transfers received exceed taxes and contributions in each of the first seven deciles (by €20,800 per CU in the first), and the balance tips at decile D8: in the last three, it is the reverse, and the top decile pays in, on average, €58,400 per CU more than it receives.
A group average does not describe its members. According to INSEE, counting imputed public services, 56% of people are net beneficiaries in 2023, 99% in the first decile and 18% in the top one. But they are a majority only in the first five deciles: in decile D7, which receives on average €1,800 per CU more than it pays in, they are only 36%. The sign of an average does not reveal the situation of the majority. A group can receive more than it pays in on average while most of its members pay in more than they receive: it is enough for the cumulated net gain of net beneficiaries to outweigh the cumulated net loss of the others. The grouped data do not say how these balances are distributed.
Receiving more than one pays in, on average, therefore offers no protection from an adjustment: position in today’s redistribution and exposure to tomorrow’s decision are two distinct rankings.
Conventions: the balance of one year, not of a lifetime; public services imputed, not measured; redistribution as a whole, not the specific effect of the debt
This is the balance of one year, not of a lifetime: retirement pensions count as current benefits. Transfers include public services — health care, education, collective services — valued in euros and allocated by statistical imputation, not a record of the services each person used; the share of net beneficiaries depends on these conventions. Income “before transfers” is a step in the calculation, not the income each person would have in an economy without taxes or public services. And this balance describes redistribution as a whole, not the specific effect of the debt.
By household age: a snapshot of the year, not a lifetime balanceINSEE, distributional national accounts 2023 (figure 1e). Groups by average age of the adults in the household.
In 2023, under these conventions, households whose adults are on average 65 or over receive €47,800 per consumption unit in transfers — of which €28,700 in cash benefits, pensions included — and pay €14,600 in taxes and contributions, against €33,800 for households aged 50 to 64. Pensions partly correspond to entitlements earned over a career, counted here as current benefits; a household’s age is not the retirement status of each of its members, and a high amount of imputed health care also reflects greater needs. This snapshot designates no payer of tomorrow.
Through which channels can the burden be distributed?
Four mechanisms can change the burden of the debt and its distribution, and they combine. The first two belong to ordinary budget adjustment, decided every year: taxes and contributions — additional taxes and contributions, or tax cuts forgone — and spending and benefits cut, frozen or postponed. The third is not decided alone: unanticipated inflation reduces the real value of claims already issued and shifts part of the burden onto their holders and onto poorly indexed incomes. The fourth is an extreme case: restructuring imposes a first-round loss on the holders of the securities, who may pass it on to their shareholders, policyholders or savers.
A diagram of possible mechanisms: no relative weight or causal effect is measured in it.
Refinancing, for its part, renews a maturity on the terms of the day: it can lighten or increase the burden, without designating who will bear it. And set against these costs is what the debt financed: services, benefits and investment benefit households today and tomorrow, and an analysis that counted only the costs would describe only half of the balance sheet.
Anticipated or not: how inflation can transfer the burden, and what works in the opposite direction
In theory, the inflation expected when a security’s rate was set tends to be already built into the rate required at issuance, and then makes no transfer. It is the gap between realised inflation and that expectation that reduces the real value of nominal claims already issued. Two mechanisms work in the opposite direction: inflation-linked securities see their cost rise with prices, and refinancing takes place at the new rates. The net effect therefore depends on the indexed share, on the refinancing calendar and on what was anticipated — and France does not choose its inflation alone. Not to be confused with the effect of inflation on the debt ratio, which works through nominal GDP and, for its part, designates no loser. The list of channels is not exhaustive: growth, asset sales and financing conditions also play a role, and history also offers cases of financial repression, which constrains returns.
Three hypotheses to be tested
The channels say where the burden can pass; three stronger hypotheses circulate about whom it would reach. None is established by the data on this page, and each calls for its own test.
Future generations — what they inherit
Debt converts present spending into future obligations: those who come after inherit them without having taken part in the decision. But they also inherit what it financed and part of the securities themselves: the net transfer is not measured here.
What the net transfer depends on, and what net debt does not say
The next generations inherit the infrastructure, education and public assets financed, and securities that their parents hold directly or through assurance-vie. This does not necessarily neutralise the burden: some models identify a channel through which debt crowds out productive capital, which this page does not measure for France. The net transfer depends on the use, the financing, the claims passed on and the alternative chosen. Net debt as published by INSEE does not settle the matter either: it deducts only certain financial assets, not roads, human capital or the environment. The page Is public debt a burden on future generations? breaks down, in the accounts, what deficits were used for, without saying which borrowed euro paid for which expense.
Less mobile groups — a hypothesis to be tested
Households and firms that can move their income, wealth or activity can escape a levy more easily than employees, retirees or users of a local public service. Hence the hypothesis: when adjustment follows the line of least resistance, it may weigh more on those who cannot leave.
To test it: groups defined in advance, a specified adjustment, a fixed measure
Testing it requires defining these groups before observing the adjustment, specifying the adjustment studied and fixing a measure of effort; the hypothesis would be weakened by an adjustment falling mainly on other groups, or by compensation granted to the losers. Geographical mobility, mobility of the tax base and the capacity to replace a public service are three distinct properties. The territorial channel is examined on the page Are local authorities the adjustment variable of public finances?
Investments with deferred benefits — a mechanism to be established case by case
A budget constrained by debt service may encourage the postponement of spending whose benefits are distant and whose postponement is politically cheap: research, maintenance of public assets, training, the ecological transition. The mechanism is plausible, not established.
Plausible, not established: no aggregate measures an investment prevented by interest payments
It remains to be established decision by decision (which spending, postponed by which decision, for which budgetary reason), since no aggregate measures an investment prevented by interest payments. Where it is established, it compounds the transfer: the benefit that was never built adds to the obligations passed on, and appears in no stock of debt.
A real, dated decision
What these data do not allow us to establish
No statistic says, on its own, who ultimately bears the burden of French debt: the final incidence of a euro of interest depends on the adjustments chosen year after year, and on the alternative they are compared with. A current distribution of taxes and benefits describes who contributes and who receives today; it does not identify the future payers. An average per consumption unit is not an individual bill. What can be established is the effect of a specific decision — a reform, a freeze, a postponement — on groups defined in advance: it is at that level that “who pays?” receives verifiable answers.
Tested against the research literature: the direction of the spending/tax contrast recurs in consolidations actually observed; the inherited-claim argument is challenged by the crowding-out of capital in Diamond’s model
Measured here. An accounting profile: the same €10 billion adjustment distributed in proportion to what each decile already pays or receives, according to INSEE, with no behaviour or feedback effects. The literature tests the readings drawn from it, not the profile itself.
Consistent with. The direction of the contrast between spending and taxes: in fiscal consolidations actually observed in 18 industrialised countries, France included, from 1978 to 2009, net income inequality rises; spending cuts appear unfavourable, tax rises equalising, although the latter effect is not significant in the baseline specification (Agnello and Sousa). The postponement of investment: in Breunig and Busemeyer’s sample, a heavier interest burden goes with a decline in the share of public investment, among the three items studied, in favour of pensions; the authors explain it by a lower political cost: postponing an investment costs less than cutting an entitlement. The role of use: in Diamond’s model, borrowing to acquire physical capital makes the government a mere intermediary between savers and entrepreneurs, with no effect.
Challenged by. Any reading in which the inherited claim would neutralise the burden: in the efficient case of Diamond’s model, where the interest rate exceeds population growth, domestically held debt lowers long-run welfare more than external debt, because it takes the place of productive capital in people’s wealth. Barro, who defends the opposite view, bases it on voluntary transfers between parents and children, operative for most people, and acknowledges that in 1989 most economists still leaned towards the standard models. The magnitude: in Agnello and Sousa’s breakdown, it is the largest tax rises that are associated with a significant fall in inequality, and the effects of consolidations tend to disappear within two years. The “tax” curve read as final incidence: in 55,082 firms in nine European countries, from 1996 to 2003, about half of a rise in corporate income tax falls in the long run on wages (Arulampalam, Devereux and Maffini) — a different tax from the one on this page, but the same gap between legal incidence (who is legally liable for the tax) and final incidence.
Not established. That less mobile groups bear more of the adjustment: the corporate tax study measures wage bargaining, and its only mobility contrast, between multinationals and independent firms, is not significant. That only unanticipated inflation transfers the burden: we did not find, in the texts examined, a test separating anticipated from unanticipated inflation. Finally, the page’s “education” exposure has no observed equivalent: a cash income indicator does not directly value a loss of public service.
References read
Agnello, L. and Sousa, R. M., “How Does Fiscal Consolidation Impact on Income Inequality?”, Banque de France working paper no. 382, 2012 (published in the Review of Income and Wealth, 60(4), 2014).
Breunig, C. and Busemeyer, M. R. (2012), “Fiscal austerity and the trade-off between public investment and social spending”, Journal of European Public Policy, 19(6), pp. 921-938.
Arulampalam, W., Devereux, M. P. and Maffini, G., “The Direct Incidence of Corporate Income Tax on Wages”, revised manuscript, February 2012 (published in the European Economic Review, 56(6), 2012).
Diamond, P. A. (1965), “National Debt in a Neoclassical Growth Model”, American Economic Review, 55(5), pp. 1126-1150.
Barro, R. J. (1989), “The Ricardian Approach to Budget Deficits”, Journal of Economic Perspectives, 3(2), pp. 37-54.
Literature review checked on 30 September 2026: each reference was read in the version indicated, in the sections on which what is said of it rests.
Cancelling the debt held by the Banque de France: who would pay?
In August 2026, Jean-Luc Mélenchon revived, for the government bonds held by the Banque de France, a proposal first made in 2020 at the Institut Rousseau for the public debt held by the Eurosystem: cancelling it. Its supporters argue that no private holder would suffer a loss, since these bonds have already been bought back; their critics dispute that it would cost the community nothing. Olivier Blanchard considers the operation has no effect: the State would save interest but lose as much in profits passed on by its central bank (as reported by Euronews, 27 August 2026). Christine Lagarde, President of the ECB, replied on 10 September that cancellation of debt “is covered under Article 123” of the Treaty, and that anything going against it “would be a pure violation of the Treaty”. One question decides between them: what would it take for cancelled debt to become a saving for the community? To answer it, we first need to see how the State borrows, and where the Banque de France sits in that circuit.
The path of a government bond, then the interest paid each year, and to whom. The numbers refer to the steps described under the diagram.
What cancellation would erase. The Banque de France reports holding, on 30 June 2026, 488 billion euros of public debt bought for the Eurosystem, at amortised cost (the purchase price, brought closer to the redemption value over time), against 546 on 31 December 2025: the amount falls at each maturity, because repaid bonds are no longer replaced. Relative to GDP, this amount represents about 16 percentage points. The actual fall in Maastricht debt (general government gross debt, excluding debt between government units, at nominal value) would depend on which French bonds were cancelled and on their nominal value, which this figure does not give, and would require the operation to be allowed and recognised by European statistics.
What it would not erase. The Banque de France paid for these bonds with money it created and credited to banks’ accounts with it: these are reserves (step 4 of the diagram). Cancelling the bonds does not remove these reserves. And they can carry a cost: the Banque de France pays the rate set by the ECB, known as the deposit rate (2.50% since 16 September 2026), on reserves that banks place in the deposit facility; this rate is the ECB’s main monetary policy instrument. Minimum reserves (the amount each bank is required to hold on its current account at the Banque de France) have earned nothing since September 2023; since June 2026, excess reserves that banks with access to the deposit facility leave on their current accounts have been remunerated at the same rate. When it bought the bonds, the Banque de France exchanged reserves for government bonds; cancellation would make that exchange permanent.
Two paths. Without cancellation, each bond that matures is repaid, if need be by new borrowing on the market; other things being equal, that repayment removes the same amount of reserves. With cancellation, the State no longer has that amount to repay or refinance, but the reserves remain. As long as a bond is running, the interest the State would stop paying is income the Banque de France would lose, even if the two effects do not necessarily hit the budget in the same year: what it passes on to the State depends on its results.
A simulation, not an observed saving. At maturity, everything comes down to one comparison: is borrowing again on the market cheaper or dearer than keeping the reserves? In 2025, keeping the reserves would have cost less; in 2023 and 2024, more. To put a figure on it, suppose each euro that matures is refinanced over ten years, and that cancellation keeps one euro of reserves paid at the deposit rate instead. In 2025, the market yield on ten-year government bonds averaged 3.35%, and the deposit rate 2.26% on average: the differential would amount to a gain of about 1.1 billion a year per 100 billion concerned, or 5.3 billion if all 488 billion were. In 2023 and 2024 the sign was the opposite: the deposit rate (3.31 then 3.73%) exceeded the ten-year rate (2.99 then 2.97%), a loss of 0.3 then 0.8 billion a year per 100 billion. The result can therefore change sign: cancellation guarantees neither a saving nor a loss. These figures are neither a forecast nor an achieved saving: they apply each year’s rates to the June 2026 amount, while the State also borrows at shorter maturities.
Who would bear the burden. Cancellation would change how burdens and income are shared between the State, the Banque de France, banks and, possibly, their customers; the result would depend on the rules for remunerating reserves and on the monetary instruments used.
Scenario
Who bears the burden
Consequence
No cancellation
The State pays the interest and repays the bonds, if need be by new borrowing; the Banque de France bears the cost of reserves until maturity
Flows follow maturities, rates and the Banque de France’s results
Cancellation, nothing else changes
The State no longer pays this interest or repays these bonds; the Banque de France loses this income and keeps the reserves
The result for the whole depends on the gap between the refinancing avoided and the cost of reserves
Cancellation and lower remuneration of reserves, as the Institut Rousseau proposes
Banks lose part of the interest they would have received; the effect on their customers and shareholders depends on their behaviour
A saving for the public sector, matched by a redistribution and a change in the monetary policy instrument
Cancellation, then recapitalisation of the Banque de France by the State
The State transfers resources to its central bank
The apparent gain is offset, partly or fully
Cancellation and a lastingly too accommodative monetary policy
The community, through inflation
A conditional risk, not an automatic consequence (objection by John Cochrane, The Grumpy Economist, 24 August 2026)
What each side gets right. Cancelling 488 billion of debt does not mean saving 488 billion. Supporters of cancellation are right on one point: no private holder would lose on the cancelled bonds. So is Olivier Blanchard, as long as the bonds are running: what the State would stop paying, the Banque de France would stop receiving. But debt disappearing from the State’s accounts does not make its cost disappear for the community, and reserves that remain do not rule out any saving. Cancellation can change the official amount of debt, how burdens are shared and when; it guarantees neither a saving nor the absence of a cost.
Law and statistics. Article 123 of the Treaty forbids the ECB and national central banks to lend to governments and to buy their bonds directly from them. It does not name cancellation: linking cancellation to it is the ECB’s reading, which the Institut Rousseau disputes by invoking the purpose of the rule, and the Court of Justice of the EU has not ruled. In national accounts, a cancellation agreed between the Banque de France and the State would, under the general rules of ESA 2010 (point 4.165 f) and subject to the decision Eurostat would take on this case, be recorded as a capital transfer from one to the other: it could improve the measured public balance once, without bringing in a single euro of cash. A refusal to pay decided by the State alone would not be recorded, and the debt would remain due. As of 18 November 2026, none of the methodological decisions published by Eurostat deals with this case.
What this comparison does not say
The State and the Banque de France taken together are an analytical perimeter, not an official statistic: the Banque de France is not part of general government. The year’s average deposit rate approximates the cost of reserves without measuring it bond by bond, and the link between a cancelled bond and a euro of reserves kept is a simplification: reserves also depend on operations across the whole euro area, and the Eurosystem has other ways than selling bonds to drain liquidity. In the public sector purchase programmes, 20% of purchases are subject to risk sharing across the Eurosystem; the French government bonds bought by the Banque de France are not subject to it. The Banque de France’s results do not depend only on its bonds and reserves: in 2025, an exceptional operation on its gold made them positive. Finally, neither the effect on market rates, nor the effect on confidence, nor the risk of inflation can be quantified with these data.
Summary
What to take away
“Who pays?” has one answer per decision, and the choice matters most at the ends of the scale. In this page’s accounting profile, the same adjustment weighs mostly on the least well-off through education, on the best-off through taxes; in the middle of the scale, the three levers studied produce similar exposures. Naming the decision and the alternative turns an opinion into a verifiable statement.
Holding is not paying, and interest can be traced only so far. Direct holders are observed, indirect holders approximated through household savings, but no statistic observes which households receive the interest (INSEE allocates it only by convention): the breakdown of holders tells us where the interest goes, never how it is financed.
A group average does not describe its members. A decile can receive more than it pays in on average while most of its members pay in more than they receive. A statistic by decile answers “how is the total distributed?”, not “what does a person in this decile experience?”.
Further reading in this dossier: two applications of “who pays?”
The reassuring argument is a serious one: by the simple arithmetic of debt, when the interest rate stays below growth, the debt ratio can stabilise, provided the deficit excluding interest — the primary balance — stays below a threshold that depends on the gap between the two rates and on the level of the debt. Beyond it, the ratio rises regardless. And even when stable, a debt has to be serviced every year: its financing and its adjustments distribute costs and benefits among taxpayers, users of public services, savers and generations. The useful question is then what it finances and who bears the adjustments — and it is answered configuration by configuration, not in principle.
Is public debt a burden on future generations?
Not mechanically. The next generations inherit the obligations, but also what they financed — infrastructure, education, public assets — and part of the securities themselves, held directly or through French assurance-vie savings policies (life-insurance-based savings) — which does not necessarily neutralise the burden, since, in some models, debt can also crowd out productive capital. The net transfer depends on the use, the financing, the claims passed on and the alternative chosen; other things being equal, debt that finances current consumption with no lasting counterpart mostly passes on a cost, and an investment they will benefit from can pass on more than it costs. What remains true in every case: the decision is taken without them.
Who holds French public debt, and who receives the interest?
Central government accounts for 82% of public debt (end of June 2026, INSEE). In Q2 2026, its marketable securities were held, at market value, 59.7% by non-residents, 21.7% by resident banks, insurers and funds, 18.7% by other resident holders — among them the Banque de France, which this breakdown does not single out (Banque de France, via Agence France Trésor). Holders are classified by residence, not nationality. For households, the French Treasury (Direction générale du Trésor), tracing the investments of assurance-vie policies, funds and banks, estimates that out of every 10 euros of financial wealth, 1 finances general government debt, mostly French. INSEE’s distributional accounts, however, allocate this interest among households only by convention, in proportion to their financial income: no published statistic observes which households receive it, and a share of the securities held is not a share of the interest paid. Holding is not paying: the holder advanced the funds and receives the return on them; the breakdown of holders does not say who bears the final burden.
Should public debt be repaid?
Each security that matures is repaid, but the debt as a whole is refinanced: the government issues new securities to repay the old ones, and the stock evolves with the deficit. Repaying the whole stock is therefore a largely theoretical hypothesis. The operative question is what happens while the debt is continuously rolled over: who bears the interest, which budgets carry the adjustments, who inherits the obligations and what they financed. Refinancing renews a maturity on the terms of the day: it can lighten or increase the burden, and on its own it does not identify who will bear it.
Can France get out of the public debt spiral?
Rolling over loans does not, by itself, condemn debt to grow. Its weight rises when the deficit excluding interest persists and when the rate paid exceeds the economy’s growth; it can grow heavier still if lenders demand higher rates, or if adjustments weaken growth. It is not inevitable: several European countries have reduced their ratio. But the ways out are not equivalent: cutting the deficit, raising growth, lowering the cost of financing or cancelling part of the debt do not put the burden on the same groups, taxpayers, recipients of public spending, savers or banks. A cancellation, if it were allowed and recognised, would reduce the stock without removing what rebuilds it. The dossier’s page “Can the rules of the public debt game be changed?” compares these routes, among others: what each changes, who decides it and on whom it puts the burden.
Redistribution by decile and by age: Mathias André, Alexis Barrau and Thomas Renaud (INSEE), “Revenus des ménages élargis à l’ensemble de l’économie en 2023”, Insee Analysesno. 118, 16 April 2026 (figures 1c, 1e and 2a), in French.
Exposure to the same adjustment: INSEE, distributional national accounts, table CND.101, twentiles of usual standard of living, 2020-2023, base 2020, released as experimental statistics; the author’s calculation, with robustness tests published with the data.
Debt by level of government: INSEE, quarterly Maastricht debt of general government by subsector (series 010777610, 010777613, 010777625 and 010777626 of its macroeconomic database), updated 29 September 2026, end of June 2026. The book, based on the accounts for end of 2025, prints 2,823 billion for central government out of 3,460.5; this page follows the latest published data.
Holders of central-government securities: Banque de France, via Agence France Trésor, Q2 2026, values read from the Agency’s chart (archived screenshot).
Households’ indirect claims: French Treasury (Direction générale du Trésor), “Que finance l’épargne financière des ménages français ?”, Trésor-Éco no. 378, 8 January 2026.
Figures are read by generators from archived documents whose fingerprint is checked; the AFT shares, published only as an image, are read by hand from the archived capture, and the values of the press releases in the section on cancellation are transcribed, then checked by script against the archived text; the compiled data can be downloaded from the “Reuse” block below.
Where this analysis comes from
Anthropy is the hypothesis that social systems displace disorder rather than resolve it.
Applied to public finances, this hypothesis is formalised in the working paper AWP-03 — Public debt and anthropy: who really pays for disorder? (DOI: 10.5281/zenodo.24368737, open-access PDF). The working paper AWP-09 tests this reading against public accounts: the accounting distribution of the same fiscal effort across households depends on the instrument chosen, but the asymmetry between mobile and captive households lies beyond what the accounts can test, since they do not observe the former escaping.
Reusing this page
The charts, the data and a ready-made citation, under an open licence. Each chart carries its source, period and reading caveat inside the image: reused in a lecture or an article, it is not separated from what makes it readable.
Who borrows? Who holds French government securities?
The levels of government that borrow, then the holders of central-government securities: two separate scopes, with no proportional link.
Reading caveat. A: nominal value; B: market value. Holding securities does not measure the final burden.
INSEE, quarterly Maastricht debt by subsector (29 September 2026), end of June 2026 · Banque de France via Agence France Trésor (French Treasury agency), Q2 2026
Net contributors and net beneficiaries by decile, 2023
On average per CU, the lower and middle deciles receive more than they pay in, the best-off the reverse; but from the middle of the scale, most people pay in more than they receive.
Reading caveat. Averages per CU; pensions and public services valued by imputation included; the balance of one year, not of a lifetime; does not measure the specific effect of the debt.
Taxes and contributions, and public transfers, by household age, 2023
In 2023, under these conventions: what each group pays in and receives, ranked by the mean age of the household’s adults, pensions and public services included.
Reading caveat. A snapshot of one year, not the balance sheet of a generation nor of future generations; a household’s age is not the retirement status of its members.
The same €10 billion adjustment: who would bear it, depending on the decision taken?
What the same adjustment would represent for each standard-of-living decile depending on the decision taken (taxes, pensions or education), then how much that choice changes what the same group bears.
Reading caveat. An accounting exposure profile, not a simulation: no behaviour, no feedback effect. A cut in a valued public service is not a loss of cash income. The D1 point of the tax decision is excluded; the reason is given in the figure’s footer.
Insee, distributional national accounts, table CND.101 (standard-of-living twentiles, 2020-2023, base 2020) — 2023 vintage; author’s calculation
Through which channels can the burden be distributed?
Four possible mechanisms (not an exhaustive list); refinancing, which rolls the maturity over without designating a loser; and, alongside, what the debt has financed.
Reading caveat. No relative weight or causal effect is measured here.
Lalut, Stéphane (2026). “Who really pays for public debt?” INSEE, Banque de France (directly and via Agence France Trésor), ECB and Eurostat data, retrieved 5 November 2026. https://stephane-lalut.com/en/who-really-pays-public-debt/
In three sentences
How the effects of public debt are distributed depends on what it finances, how it is financed and the adjustments chosen to service it; some configurations might shift costs onto groups less able to avoid them, a hypothesis these data do not test. The available data show who borrows, who holds central-government securities and who contributes or receives today — not who will bear the final burden. A loss can only be attributed to a specific decision, compared with its alternative.
The data
Public debt by government subsector (INSEE), holders of central-government marketable securities (Banque de France via Agence France Trésor) and INSEE’s 2023 distributional national accounts, with their periods, units and conventions, and the robustness tests of the exposure profile.
Data and charts CC BY 4.0 — free reuse, including commercial, with attribution. Text under copyright — short quotations permitted, full reproduction on request. The raw series belong to their producers (INSEE, Banque de France [directly and via Agence France Trésor], ECB and Eurostat), under their own terms. Values taken from Agence France Trésor (AFT), in the files and in the charts, are not covered by the CC BY licence: that producer restricts their commercial reuse.