Wiping out the debt? What Mélenchon is really proposing, and what it would change
On 23 August, Jean-Luc Mélenchon called on the ECB to "put in the freezer" €636bn of French debt. The proposal contains a real accounting truth and an equally real economic dead end. A jargon-free breakdown: internal versus external debt, what the operation would really save, and why the real debate lies elsewhere.
On 23 August, at his summer university near Valence, Jean-Luc Mélenchon (leader of the left-wing La France insoumise) called on the European Central Bank to "put the debts of states in the freezer, starting with the Covid-era debt". The remark set off a storm. It deserves better than outrage or applause: it contains a real accounting truth and an equally real economic dead end. This article separates the two, without jargon.
What was actually proposed
First, and it is no detail: nobody proposed to cancel the debt in the strict sense.
The proposal is to convert the French government bonds held by the Banque de France (France's central bank), on behalf of the ECB, into zero-coupon perpetual debt. A perpetual debt is a loan with no repayment date: the borrower never returns the principal. At a zero rate, it pays no interest either. Economically, a zero-rate perpetual is very close to a cancelled debt, but legally and in accounting terms it is not the same operation.
The amount at stake, according to the candidate: €636bn, or about 18% of French public debt, which stood at €3,536bn at the end of the first quarter of 2026 according to Insee (France's statistics office).
Imagine a loan you never have to repay and on which you pay no interest. On paper, the debt still exists. In practice, it costs you nothing and will never be called in. That is what "putting it in the freezer" means.
Where the proposal is right
Let us start with what is correct, because it is what the sharpest critics often gloss over.
The Banque de France belongs to the French state, its sole shareholder. When it makes a distributable profit, that profit flows back to the state budget as dividends and tax: between 2015 and 2022, nearly €32bn was paid back this way.
The consequence: when the state pays interest on securities held by the Banque de France, it is paying an entity it owns. The money leaves one pocket and enters the other. This is the consolidated public sector: add the state and its central bank into a single account, and this debt and this claim cancel out.
A parent company lends money to a subsidiary it owns outright. In the group accounts the loan does not appear: it cancels itself out, because lender and borrower are the same house.
The relationship between the French state and the Banque de France resembles this. Membership of the Eurosystem complicates matters, but the principle holds.
So yes: erasing this line would not wipe out hundreds of billions of euros of savings belonging to an individual or an insurer. A saver holding a life-insurance contract would lose nothing from this particular operation. On that point, the panic is miscalibrated.
The distinction that governs the whole debate
To understand why the proposal still does not hold, we need a distinction that public debate constantly muddles: internal debt versus external debt.
The state borrows from French people (households, banks, insurers) or from its own central bank.
National wealth does not move when the loan is made. A resource is simply shifted from one French pocket to another.
If the money is badly spent, the impoverishment has already happened at the moment of spending, not at repayment.
The state borrows abroad, where it levies no tax and controls nothing.
The country gets resources from outside: a real, temporary inflow.
But at repayment, it has to give them back. That is where the cost sits, and where a refusal to pay creates a conflict.
A family in which the father lends his son €1,000 does not get poorer: the money stays in the house. If the son spends it badly, the family is poorer, but it became so on the day of the spending, not the day of repayment.
If the son borrows from the neighbour and squanders the sum, the family first lived beyond its means. The bill arrives when the neighbour wants his money back.
That is why the ownership structure matters so much. About 56% of the state's marketable debt was held by non-residents at the end of 2025, according to the Banque de France. That portion is not part of an internal bookkeeping game: these are real foreign creditors, who have no reason to accept losses in place of the French.
Mélenchon's proposal covers only the €636bn held at the Banque de France. It does not touch foreign creditors. That is coherent, and it is also what drastically limits its reach.
Where the reasoning breaks down
Here is the heart of the problem, and it fits in one sentence: you cannot say that this debt does not matter and then claim credit for erasing it in order to create new debt.
The argument runs as follows. By taking €636bn off the counter, the debt ratio would mechanically fall from about 117.5% to a little over 96% of GDP. France would then regain room to borrow again and finance the ecological transition, public services and investment.
But look closely at the logic. It is one or the other.
- Either this debt really counts. In that case it is a real constraint, and it cannot be made to vanish with a stroke of the pen without consequences.
- Or it does not count, because it cancels out in the consolidated account, which is the argument put forward. In that case erasing it frees no new room, since by construction it was already weighing on nothing.
A purely accounting operation creates no wealth. Erasing an entry on a balance sheet builds neither a factory, nor an engineer, nor a tax receipt. If the operation is neutral, which is precisely the argument used to justify it, then it releases no additional spending capacity. The advertised gain is a gain in presentation.
You cannot invoke an operation's neutrality to make it acceptable, then invoke its effects to make it useful.
And what if the tap were reopened anyway?
That leaves the fully assumed version of the approach: erase, then borrow again, with the central bank buying the new debt by creating money. That is what was done during Covid, when roughly €250bn was injected into the French economy through this channel.
The argument "we did it in 2020, so we can do it again" deserves a precise answer, and it has two parts.
First, the context was not the same. In 2020 the economy was at a standstill, demand had collapsed and inflation was nil. Creating money against weak demand does not push prices up. Today the situation is the reverse.
Second, the bill came due. Inflation in France hit 5.2% in 2022 and 4.9% in 2023. The energy shock contributed heavily, but the massive monetary expansion of 2020-2021 was not a bystander. To say that "everything went fine" is to leave that bill out of the count.
Creating money does not create goods. If the quantity of euros grows faster than the quantity of things to buy, each euro buys less. That is inflation, an invisible but real levy that hits first those whose incomes do not keep pace.
The legal obstacle, which is not a small one
Article 123 of the Treaty on the Functioning of the European Union prohibits the monetary financing of states: a central bank cannot lend directly to a government. It can, however, buy securities already issued on the market, which it did on a massive scale from 2015, then in 2020.
This distinction may seem Byzantine. It is in fact the pivot of the whole European monetary edifice: it ensures that the central bank does not become the Treasury's cash window.
Christine Lagarde, then President of the ECB, answered a similar request in 2021 by saying that a cancellation would breach the treaties, the ban on monetary financing being "one of the fundamental pillars" of the euro. Xavier Ragot, president of the French Institute of Economics, made the same objection about the August proposal: freezing debt held by the ECB contravenes treaties that guarantee its operational independence.
That reading is not unanimous. In February 2021, more than a hundred economists, Thomas Piketty among them, published an appeal calling for the cancellation of public debts held by the ECB, in exchange for an equivalent commitment to ecological and social investment. Their thesis: a cancellation would not be monetary financing within the meaning of the treaty, since the securities were already lawfully acquired on the secondary market, and the ECB is free to dispose of its balance sheet.
The legal debate is therefore real. But it remains a minority view among EU lawyers, and above all the ECB itself settles it the other way. And it is the ECB that would decide.
What it would really bring in
Now the numbers, which are the most illuminating exercise.
The debt burden, meaning the interest paid each year, is approaching €59bn in 2026. That is a considerable sum, comparable to the national education budget.
But not all of that interest is a loss for France. It has to be broken down.
| Recipient | Order of magnitude | Effect for France |
|---|---|---|
| Banque de France | Left pocket → right pocket | Returns to the state through dividends and tax. Net cost close to zero. |
| French savers, banks and insurers | Stays in France | It is public spending, but also French income. Internal transfer. |
| Foreign creditors | ≈ half of the €59bn | The only real leak of wealth out of the country. |
In other words, the only genuine saving that a full cancellation would produce, including of foreign-held debt, which nobody proposes, would be on the order of €30bn a year. Compare that with a public deficit of €152bn in 2025.
Even by wiping out all of France's debt, the deficit would not be closed. The hole comes from the state spending markedly more each year than it takes in, and no accounting entry fixes that.
What the market is already saying, without waiting for 2027
There is a little-discussed reason why states borrow on a market rather than from their own central bank: the market acts as a counterweight. It continuously assesses a borrower's soundness through two simple signals.
- Does the debt find buyers? At each auction, the Agence France Trésor (France's debt-management agency) watches how many investors show up.
- At what rate? The more a borrower worries lenders, the more it pays.
Both signals are deteriorating. The 10-year OAT yield, the benchmark French government bond, stood at 4.24% in early September 2026, after crossing the 4% threshold at the end of August. The gap with the equivalent German bond, the so-called spread, reached about 84 basis points on 21 August, or 0.84 percentage points.
The spread measures the risk premium: how much more France pays than Germany to borrow the same sum, for the same term, in the same currency. The wider it is, the less relaxed lenders are.
Demand is still there, with recent auctions heavily oversubscribed, but it is costing more and more. That is exactly the role of the market signal: it does not slam the door, it raises the toll.
The danger of a unilateral cancellation is not the loss of €636bn at a private creditor, since that would not happen. The danger is what the operation would signal: a weakened national central bank, European rules set aside, a budget policy perceived as limitless. The penalty would come not from the debt erased, but from the price of the next debt, the one that has to be issued every week to keep the state running.
The real debate is not that one
This proposal deserves credit for one thing: it raises a legitimate question. The debt constraint weighs on collective choices, it limits public investment at a time when the energy transition demands a lot of it, and the reminder that the Banque de France is not an ordinary creditor is well founded.
But it offers an accounting solution to a problem that is not an accounting one. The level of debt, taken in isolation, is not the right indicator. What determines a country's solvency is a quartet: the gap between revenue and spending, growth, the cost of borrowing, and the external balance.
The Swiss comparison shows it well. The Confederation reports net debt of about 16% of GDP, surplus public accounts, and a largely creditor external position. Its central bank holds only a tiny fraction of federal debt. Nobody knows the amount of Swiss debt, and for good reason: in that configuration it does not matter.
The French case is the opposite. For 2026 alone, the state expects about €325bn in net revenue against €459bn of spending: it spends nearly 41% more than it takes in. No operation on the existing stock of debt corrects a flow deficit of that size.
The stock is what you owe. The flow is the gap between what comes in and what goes out each month. Erasing part of what you owe changes nothing about the fact that you spend more than you earn. The hole reopens immediately.
Key takeaways
- The proposal is not absurd in accounting terms. Debt held by the Banque de France does offset within the public sector. Those predicting ruin for savers have the wrong mechanism.
- Yet it frees no room. An accounting entry creates no resource. If the operation is neutral, the very argument of its promoters, it releases nothing.
- It runs into a serious legal obstacle, contested by a minority of economists, but settled the other way by the ECB itself, which is the deciding authority.
- The real saving would be limited: only interest leaving for abroad is a leak, about €30bn a year against a deficit of €152bn.
- The potential cost, by contrast, would be immediate: the French risk premium is already tightening without any such decision having been taken.
- The underlying issue remains untouched. A country does not become solvent by reorganising its entries, but by bringing spending closer to revenue or by producing more. That is where the 2027 debate should take place.
Glossary
A loan with no repayment date. The principal is never returned; only interest is paid, and not even that if the rate is zero.
An overall view grouping the state and the bodies it owns, including its central bank. Debts internal to that perimeter cancel out within it.
A central bank creating money to finance its government directly. Prohibited by Article 123 of the European treaty.
The market where already-issued securities are traded between investors. The ECB buys there, which is what makes its purchases lawful.
Obligation assimilable du Trésor: the French state's main long-term debt security. The 10-year OAT serves as the benchmark rate.
The yield gap between two comparable borrowers. The OAT-Bund spread measures France's risk premium relative to Germany.
Total interest paid in the year. About €59bn expected in 2026. Does not include repayment of principal.
Methodology note and sources
This article starts from a video by the channel Grand Angle on the subject, whose transcript was extracted and whose economic reasoning (the internal/external distinction, the consolidation of the public sector, the role of the market as a counterweight) served as the guiding thread. The figures given in that source were, however, rechecked and corrected: the amount of debt held at the Banque de France was given as between €469bn and €550bn, whereas the proposal covers €636bn; and the proposal was presented there as a cancellation, when it is a conversion into zero-rate perpetual debt. The case made by supporters of cancellation, absent from the source, was added for balance. This article aims to explain a mechanism, not to take sides in the 2027 campaign.
- Insee — public debt at 117.5% of GDP in Q1 2026
- Insee — 2025 public deficit at 5.1% of GDP
- Franceinfo — five questions on the proposal
- AFP / Boursorama — criticism of the proposal
- Le Temps — feasibility analysis
- The economists' appeal of February 2021
- Le Grand Continent — the legal analysis
- ECB — official reply on cancellation
- Senate — ownership structure of state debt
- Agence France Trésor — auctions and outstanding debt
- OAT-Bund spread — tracker
- Transcript: Grand Angle, "Why Mélenchon wants to cancel French debt", 30 August 2026.