What really happens when a country stops paying its debt?

On 12 December 2008, Rafael Correa announced that Ecuador would not pay roughly $31 million of interest due on an international bond. The gesture was spectacular. The following year, Quito repurchased 91% of the defaulted securities after setting a price of 35 cents per dollar of face value. It looked as though $3.2 billion of debt had vanished with the stroke of a pen. Six years later, however, the country returned to the market to borrow $2 billion at 7.95%. Between those dates lies the entire mechanism of a sovereign default: debt can be reduced, but the loss never disappears. It changes balance sheet, calendar and sometimes victim.

Ecuador’s story feeds two opposite narratives. To some, the episode establishes that a state can refuse to pay without disaster. To others, every default condemns a country to immediate ruin. Both narratives omit the same questions: who held the debt, in which currency, what still had to be imported, who could refinance the state, and what happened before the missed payment?

Five points frame the problem:

1. A state does not enter a global bankruptcy court. It continues to tax, pay civil servants and pass laws.
2. Default does not necessarily cover all its debt. It can concern one bond, external creditors or domestic debt.
3. The reduction obtained is real for the state, but it is somebody else’s loss.
4. The economic crisis often begins before default. The missed payment can worsen it, shorten it or simply acknowledge it.
5. The difference between an orderly exit and a disaster depends less on the word “cancellation” than on transition financing, the banking system, the currency and the speed of the agreement.

Missed payment, moratorium, restructuring, haircut and cancellation

A missed payment occurs when an amount due is not paid. Some contracts include a grace period before the default becomes legally effective.

A moratorium is an announced suspension of payments. It buys time, but by itself changes neither the amount owed nor the contract.

A sovereign debt restructuring changes the contract: longer maturity, lower coupon, grace period, principal reduction, or a combination of these terms. A coercive exchange that imposes a loss on creditors can qualify as a default even when no coupon was missed.

A sovereign haircut measures the creditor’s economic loss. It is not always the same as the reduction in face value. Receiving 100 in thirty years instead of tomorrow already produces a loss in present-value terms.

A cancellation legally extinguishes a claim. It can be bilateral, multilateral or part of a comprehensive agreement. But it does not erase the bank reserves created when a central bank bought the bond, the hole left in a pension fund, or the need to finance the next budget deficit.

Modern bonds often include a collective action clause, or CAC. A qualified majority can approve an exchange that binds minority holders. The clause limits holdouts, but it does not create a global bankruptcy code. The operational playbook published in 2025 by the Global Sovereign Debt Roundtable therefore describes a negotiation: debt inventory, sustainability analysis, perimeter, creditor committee, agreement, exchange and new financing.

The crisis usually starts before default

Saying that “default created poverty” can be as misleading as saying that “cancellation created the recovery”. A state usually stops paying because output is falling, tax receipts are shrinking, foreign-exchange reserves are exhausted or refinancing rates have become impossible. Default is therefore endogenous to the crisis: it is simultaneously a symptom, a distributional choice and sometimes an accelerator.

That does not make default costless. A causal study of Argentina published in the American Economic Review uses court rulings that changed the probability of another default in 2014. The authors estimate that a ten-percentage-point rise in default probability reduced the value of Argentine equities by about 6% and weakened the exchange rate by about 1%. The shock reached companies with no direct reliance on the state: sovereign risk contaminated the private economy. Hébert and Schreger, 2017

Historical comparisons point in the same direction, with an essential caveat. A 2024 working paper studies 221 external defaults between 1815 and 2020 using synthetic control groups. It estimates that real GDP per capita in defaulting countries stood 8.5% below peers after three years and almost 20% below that counterfactual after ten. The number of poor households may have been about 10% higher at ten years, although poverty data are scarcer and less robust. The authors themselves note that crises often begin before default. This is a counterfactual trajectory estimate; it does not establish that the legal act alone caused every loss. Farah-Yacoub, Graf von Luckner and Reinhart, NBER, 2024

Peer-reviewed work does not produce one universal magnitude. Kuvshinov and Zimmermann estimate an output gap of 2.7% at default and 3.7% at its peak five years later, followed by longer-run catch-up. Their estimate reaches about 9.5% when default is followed by a systemic banking crisis. These results validate neither automatic apocalypse nor harmlessness: the sovereign-bank loop changes the order of magnitude. Kuvshinov and Zimmermann, European Economic Review, 2019

Average GDP also hides distribution. A study of 124 developing countries between 1980 and 2016 associates sovereign default with greater income and wealth inequality that can persist for five years after the episode ends. The mechanisms include less redistributive capacity through lower taxes, subsidies and social spending. The estimate varies with the size, length and type of default and with institutional quality. Apeti, Industrial and Corporate Change, 2023

Two further findings resist caricature. Across 180 restructurings from 1970 to 2010, deeper haircuts are associated with higher later spreads and longer market exclusion. “Hard” defaults, involving more conflict and larger creditor losses, are followed by higher output costs than cooperative restructurings. Cruces and Trebesch, 2013, Trebesch and Zabel, 2017

Timing can be measured too. Asonuma and Trebesch identify 38% of the 1978-2010 restructurings as pre-emptive. On average, they were quicker, involved smaller haircuts and were followed by lower output losses than post-default restructurings. Later research in the Journal of International Economics also finds worse outcomes after default, particularly where the private sector relies heavily on domestic bank credit. These are empirical associations and a quantitative model, not a guarantee that acting early is sufficient. They do reject the idea that waiting for the last dollar is always prudent. Asonuma and Trebesch, Journal of the European Economic Association, 2016, Asonuma, Chamon, Erce and Sasahara, Journal of International Economics, vol. 152

The evidence does not say “never restructure”. Continuing to service unsustainable debt can destroy more output and delay settlement, as the Greek programme evaluation later acknowledged. It says instead: timing, method and perimeter matter.

Ecuador: the targeted repurchase of two Eurobonds

Rafael Correa’s decision was real, but far narrower than the legend. Ecuador suspended service on the Global 2012 and Global 2030 bonds, worth about $3.2 billion. It continued to service a third issue, Global 2015, and obligations to other creditor categories. The IMF database classifies the 2008-09 operation as a post-default buyback of two Eurobonds, with an estimated economic haircut of 68.6%. IMF, restructuring history

In 2009, the government repurchased about $2.9 billion of face value for nearly $900 million. The World Bank estimates the interest saving at roughly $300 million a year. That fiscal gain was tangible. World Bank, Ecuador Public Finance Review, 2019

Three qualifications change its interpretation.

First, Ecuador was dollarised. It could neither print the currency in which its debt and imports were denominated nor devalue a national currency to restore competitiveness. Adjustment depended on oil revenue, taxes, spending and external finance.

Second, the country had an exportable asset. The World Bank notes that strong growth from 2001 to 2014 benefited from favourable external conditions and high oil prices. When those prices reversed in 2014, the vulnerability returned. World Bank, Public Finance Review

Third, access to finance did not disappear; it changed source and price. Quito relied more heavily on China and oil-linked lending during its time away from bond markets. In June 2014, it returned to sell $2 billion of ten-year bonds at a 7.95% yield, a high cost in that era of very low global rates. Reuters on Chinese financing, Bloomberg on the 2014 issue

Consumption poverty nevertheless fell sharply, from 38.3% in 2006 to 25.8% in 2014, and extreme poverty from 12.9% to 5.7%, according to the joint report by Ecuador’s statistical institute and the World Bank. Those figures cover the whole period, before and after default. The report links them to growth, employment, wages and social policy. It cannot isolate a causal contribution from the $300 million in annual interest savings. INEC and World Bank, 2016

Ecuador’s buyback did reduce external commercial debt and create fiscal room. It does not allow the conclusion that a generalised default would have produced the same fall in poverty without oil, alternative lenders and a very precise selection of targeted bonds.

Very different operations behind the word “cancellation”

Case	What actually happened	Main shock or protection	Exit route
Ecuador	Two external bonds repurchased at a deep discount	Targeted debt, oil and alternative lenders available	Market return in 2014 at 7.95%
West Germany	Multilateral 1953 agreement cutting roughly half of external debt	Restored credit, protected investment and export capacity	Growth and reintegration
Argentina	Default after three recession years, followed by devaluation	Frozen deposits, inflation, poverty and litigation	Strong rebound but long exclusion
Greece	2012 private-sector exchange cutting debt by €107bn	Bank losses followed by public recapitalisation	Private claims became official loans
Russia	Domestic debt restructuring, moratorium and rouble devaluation	Insolvent banks and losses to incomes and savings	Rebound through FX, capacity and oil

Germany in 1953: the conditions for successful relief

The London Agreement of 27 February 1953 is the strongest historical case for debt relief supporting prosperity. Its architecture also identifies the conditions that separate successful relief from a solitary refusal to pay.

The agreement brought West Germany together with public and private creditors. It settled pre-war and post-war external debt, reduced roughly half of it, and adapted the remainder to the country’s capacity to transfer resources. The text expressly considered Germany’s economic position and the objective of avoiding economic dislocation. Agreement registered with the United Nations

A study in the European Review of Economic History associates the agreement with lower borrowing costs, more fiscal room for investment and social expenditure, and macroeconomic stabilisation. Using Bundesbank monthly reports and before-and-after comparisons, its authors conclude that debt relief contributed to growth. They do not claim it was the only driver. Galofré-Vilà, McKee, Meissner and Stuckler, 2019

The setting was exceptional: productive reconstruction, American aid, Allied security, external demand, Western anchoring and a political determination to reintegrate Germany into trade and credit. Creditors did not merely surrender claims. They helped restore the export revenues from which the balance could be paid.

The London Agreement combined four conditions: cut debt deeply enough, protect investment, allow the debtor to earn foreign currency and reopen financing. Its success cannot be separated from that package.

Argentina in 2001: from recession to default

When Argentina defaulted in December 2001, the economy was already in its third year of recession. Convertibility, the system that fixed one peso to one dollar, ended only in January 2002. The IMF’s independent evaluation therefore describes a sequence rather than an isolated shock: an overvalued exchange rate, rising debt, capital flight, prolonged international support, deposit restrictions, default, devaluation and inflation. IMF Independent Evaluation Office, 2004

The social cost was immense. The World Bank estimates that poverty rose from 37% in 2001 to a peak of 58% at the end of 2002, while the number of indigent people doubled. Its report does not assign that increase to default alone. It jointly cites the collapse of convertibility, deposit restrictions, external non-payment, inflation, falling output and devaluation. World Bank, Argentina Crisis and Poverty 2003

The rapid recovery was also a package. Devaluation restored competitiveness, idle capacity allowed output to return without waiting for new investment, commodity prices supported exports, and suspended debt service released resources. But savers absorbed part of the adjustment through conversion and the real loss on deposits. Creditors took deep haircuts. Litigation with holdouts complicated a normal return to markets for years.

Argentina illustrates how default can end an impossible path and precede recovery. It also reminds us how quickly wages, savings and poverty can carry the bill. Our analysis of the IMF’s exceptional exposure to Argentina documents why access to stable financing remains central twenty-five years later.

Greece in 2012: when private losses return to the public balance sheet

Greece is Europe’s best laboratory for the question “who pays?”. In March 2012, about 97% of roughly €197 billion in privately held bonds were exchanged. Face value was cut by 53.5%, reducing the debt stock by close to €107 billion. European Stability Mechanism

For the state, the reduction was enormous. For Greek banks, it was a capital loss. The Bank of Greece estimates losses of €37.7 billion on their government bonds and loans, a major reason for a recapitalisation programme envelope of €50 billion. Bank of Greece, December 2012 report

Part of the reduced private debt was thus replaced with official loans used to stabilise the state and its banks. The ESM finds that the debt ratio fell only briefly, while longer maturities and lower rates greatly reduced annual financing needs. The distinction matters: a restructuring can ease payment flows without eliminating as much net debt as the headline suggests. ESM, retrospective study of PSI

The IMF’s ex-post evaluation acknowledges that the 2010 programme neither restored confidence nor prevented a far deeper recession than forecast. Banks had lost 30% of deposits and restructuring, delayed because of contagion fears, eventually became unavoidable. The OECD measures a 26% fall in real GDP during the depression. That collapse encompassed austerity, a banking crisis, the impossibility of devaluation within the euro and delayed restructuring. It cannot be attributed to the March 2012 exchange alone. IMF, ex-post evaluation, OECD, 2016 survey

Russia in 1998: a rapid exit can still be socially brutal

On 17 August 1998, Russia devalued the rouble, restructured its short-term domestic debt and imposed a moratorium on some private external payments. The IMF describes a de facto default on domestic debt combined with a large devaluation. Much of the banking system became formally insolvent. IMF, Russia Rebounds, banking chapter

Poverty peaked around 40% in 1999 according to Goskomstat data reported by the World Bank. Average growth then reached 6.4% a year from 1999 to 2002. The same report attributes the rebound to import substitution after devaluation, spare industrial capacity, higher energy prices, stronger tax receipts and macroeconomic stabilisation. World Bank, Russia Development Policy Review, 2003

Two propositions can therefore be true at once. The 1998 rupture ended an exchange-rate and debt regime that could not survive. It also destroyed real income, savings and banks before devaluation and oil supported the recovery. Saying only that “Russia bounced after default” omits the households that paid in between.

Are bad outcomes really more numerous?

Modern history contains many painful cases. But listing poor countries after default is not enough to attribute their poverty to default. Rich, stable states rarely default. States hit by war, banking collapse, capital flight or a balance-of-payments crisis do so more often. The sample is selected against them from the outset.

Lebanon illustrates the trap. The World Bank dates the start of its financial crisis to the halt in capital inflows in October 2019, before the country’s first sovereign default in March 2020. The banking crisis, multiple exchange rates, the pandemic and the Beirut port explosion then formed what the Bank called a “deliberate depression”, with a strong emphasis on policy inaction. Default did not repair the system because it was not quickly followed by a comprehensive allocation of banking and state losses. World Bank, Lebanon Economic Monitor, autumn 2020

In Sri Lanka, the government suspended external payments in April 2022 when reserves were already too low to finance fuel, medicine and other imports normally. The World Bank estimates that poverty at the $3.65-a-day threshold in 2017 purchasing-power terms rose from 13.1% in 2021 to 25.6% in 2022, adding 2.7 million people. It attributes the break to the economic crisis, shortages, inflation and collapsing incomes, not to one legal event. It also warned that a slow restructuring would prolong the crisis. World Bank, Sri Lanka Development Update, October 2022

By contrast, the HIPC and MDRI initiatives organised coordinated debt relief for heavily indebted poor countries. In 2016, the IMF counted 36 countries at completion point, a sharp decline in debt service and an increase in poverty-reducing expenditure. A Fund staff study finds a possible positive growth effect, while explicitly warning that relief is hard to separate from accompanying financing, reform and institutional change. IMF, HIPC-MDRI update, Marcelino and Hakobyan, 2014

The literature also finds a difference between private and official creditors. Using synthetic controls for 23 cases from 1970 to 2017, Marchesi and Masi find persistent output losses after defaults to private creditors but no comparable permanent decline following agreements with official creditors. The sample is small and cannot turn that distinction into a universal law. The quality of exit financing nevertheless matters as much as the share of debt removed. Marchesi and Masi, Journal of International Money and Finance, 2021

Coordinated relief accompanied by new financing and a growth strategy has a better record than late default in the middle of a banking and currency crisis.

How does the loss travel through the economy?

The first payer is visible: the creditor receives less or later. Four transfers can then blur that simplicity.

1. If a domestic bank holds the debt, the loss reduces its capital. The state may have to recapitalise it, as Greece did. Public debt disappears on one side and returns as bank support on the other.
2. If a pension fund holds the bond, fully protecting pensioners reduces the available haircut or shifts the charge to the budget. Failing to protect them turns restructuring into a loss of wealth or benefits.
3. If debt is denominated in foreign currency, stopping payments preserves dollars in the short run, but lost financing can make fuel, medicine and production inputs harder to import.
4. If the state still runs a primary deficit, it has to borrow again the day after default. Without a transition lender, it cuts spending, raises taxes, accumulates arrears or creates money where it can.

Academic work explains why the loss extends beyond the bond portfolio. Gennaioli, Martin and Rossi describe how banks’ exposure to their own sovereign can transmit default to private credit and force authorities to choose among austerity, inflation and recapitalisation. Sandleris models a contraction in domestic credit even when local agents do not directly hold sovereign bonds, because financing access and resource reallocation deteriorate. Gennaioli, Martin and Rossi, Journal of Finance, 2014, Sandleris, Journal of Money, Credit and Banking, 2014

For economies dependent on imported inputs, Mendoza and Yue formalise another channel: exclusion from credit forces companies to replace some working-capital-financed inputs with less efficient substitutes. Their model does not directly measure each historical episode. It clarifies how a shortage of foreign currency or trade credit can reduce output before factories or workers have physically disappeared. Mendoza and Yue, Quarterly Journal of Economics, 2012

Reading only the haircut percentage is therefore like reading one side of a bank transaction and ignoring its counterparties.

Why does a central bank buy government debt?

One vocabulary correction prevents a great deal of confusion: a central bank does not buy back “its own debt” when it purchases government bonds. The bonds are liabilities of the Treasury. They become assets of the central bank. Its own liabilities are mainly banknotes and the reserves that commercial banks hold with it.

The visible action is similar in every case: the central bank buys a security in the market and credits reserves in exchange. The objective, maturity, intervention rule and expected holding period identify the policy.

1. Managing bank reserves and short-term rates

Banks settle payments with central-bank reserves. The central bank must supply an amount consistent with its framework for controlling overnight rates. It may therefore buy short-term securities because demand for banknotes and reserves grows with the economy, or because the Treasury’s account temporarily drains reserves from the banking system.

That is the purpose of the Fed’s reserve management purchases, or RMPs, begun in December 2025. Vice Chair Philip Jefferson stressed the distinction in January 2026: bill purchases maintain “ample” reserves and control overnight rates; they do not aim to lower long yields or change the stance of monetary policy. Federal Reserve, speech of 16 January 2026

2. Easing financial conditions through QE

When the policy rate is near its lower bound, a central bank can buy large quantities of longer-dated bonds. By removing duration and sometimes credit risk from the market, it seeks to reduce long-term yields, support asset prices, ease credit and stimulate demand. That is quantitative easing, or QE.

The balance sheet grows in the previous case too, but the intended channel differs. The Fed’s distinction is clear: QE removes duration risk to affect broad financial conditions; reserve management purchases concentrate on short securities to implement the policy rate already chosen. Balance-sheet size alone does not identify the policy.

3. Repairing a market that has stopped functioning

A central bank may buy bonds not to stimulate the economy but to break a forced-selling spiral. In September 2022, the Bank of England temporarily bought long gilts when margin calls on pension funds threatened financial stability. It maintained its objective of reducing the monetary-policy portfolio, then sold the £19.3 billion acquired during the intervention. Bank of England, 28 September 2022 announcement, case study and sales

The instrument, buying a bond, resembles QE. The function is different: restore trading and give leveraged institutions time to reduce risk.

4. Fixing a yield through YCC

Yield curve control, or YCC, reverses the quantitative logic. Under QE, the central bank generally announces a purchase amount and allows the yield to adjust. Under YCC, it announces a yield or band at a maturity and promises to buy the quantity required to defend that target.

The United States actually used it from 1942 to 1951. At Treasury’s request, the Fed fixed the Treasury-bill yield at 0.375% and implicitly capped long-bond yields at 2.5% to facilitate war finance. The quantity of securities and money became the adjusting variable. The 1951 Treasury-Fed Accord restored the central bank’s autonomy on this point. Federal Reserve History, The Treasury-Fed Accord

The Bank of Japan applied YCC around the ten-year JGB yield from 2016 to March 2024. The Reserve Bank of Australia targeted the three-year Australian government bond from March 2020 to November 2021. Its retrospective review concludes that the target lowered funding costs, but a calendar-based commitment adjusted poorly to a faster-than-expected recovery. BoJ, September 2016 framework, BoJ, March 2024 change, RBA, yield-target review

The United States is not operating YCC in August 2026. The Fed has announced no ceiling for the 30-year Treasury yield. Its reserve management purchases cover Treasury bills and, if necessary, securities with no more than three years of remaining maturity. Their amount depends on reserve demand and money-market conditions, not a long-yield target. Federal Reserve Bank of New York, current terms

5. Preventing fragmentation in a currency union

In the euro area, a disorderly rise in one country’s spread can prevent one policy-rate decision from reaching every member. The 2012 OMT framework and the 2022 TPI can therefore permit targeted secondary-market purchases under different conditions. OMT requires a European programme with conditionality. TPI eligibility considers compliance with the fiscal framework, the absence of severe macroeconomic imbalances, debt sustainability and sound macroeconomic policies. ECB, OMT technical features, ECB, TPI

Neither is a permanent entitlement to financing nor a promise to set the yield preferred by a national government. Their stated purpose is the transmission of a common monetary policy.

6. Reinvesting repayments from an existing portfolio

When a bond held by a central bank matures, the Treasury repays principal. The bank can allow its balance sheet to shrink or use the amount to buy another security. In the second case it reinvests. The stock of assets may remain unchanged without new net easing.

This can create the impression that the debt is never repaid, but two transactions occur: the first security is paid, then another is bought at the market price. The issuer remains liable on the new bond and the central bank retains the option to stop reinvestments.

A consolidated view of the public balance sheet

Consider a government bond with a value of 100 held by a commercial bank. The central bank buys it and credits 100 in reserves:

1. the Treasury still owes the bond’s interest and principal;
2. the central bank owns an asset worth 100;
3. the commercial bank owns 100 in reserves;
4. the central bank owes those 100 in reserves to the bank, often remunerated at the policy rate.

If Treasury and central bank are consolidated, the government bond held inside the public sector disappears from the consolidated accounts, but the reserves held by the private sector remain. The operation has mainly replaced long-term fixed-rate debt with a very short monetary liability whose cost resets with the policy rate. Interest-rate risk has not vanished; it has moved onto the public balance sheet. The U.S. Treasury Borrowing Advisory Committee itself describes reserves as overnight floating-rate liabilities in this consolidated view. Treasury Borrowing Advisory Committee, February 2026

Cancelling the bond would remove the central bank’s asset, not the reserves. The loss would reduce future profit, remittances to the Treasury or central-bank equity. A central bank can operate through losses, but that does not turn the loss into a free resource. In a consolidated public-balance-sheet model, Ricardo Reis formalises QE as an exchange of one public liability for another, with no automatic change in state solvency. Reis, NBER Working Paper 22415, 2016

The politically interesting boundary appears when purchases become a permanent promise to maintain one government’s financing cost regardless of inflation, or when the central bank waives repayment. The policy then moves from a reversible monetary instrument towards fiscal financing. That choice can be defended politically. It cannot honestly be described as disappearance without a counterpart.

Mélenchon’s exact words on 23 August 2026

At his closing rally at the AMFiS summer gathering, Jean-Luc Mélenchon argued that the ECB could and should, in his words, “mettre au congélateur les dettes des États”, meaning put government debts “in the freezer”, beginning with pandemic-era debt. He then said that the United States had just done the same “for billions and billions”. The passage starts at 38 minutes 37 seconds in the full video published on his official YouTube channel.

The claim often paraphrased as “the United States cancelled its debt” is therefore not an exact quotation. Mélenchon spoke of freezing debt. But his comparison with the American operation is still factually wrong.

Washington’s announcement

On 19 August, the U.S. Treasury announced that from 9 September 2026 it would raise from a maximum of $2 billion to at least $4 billion the maximum size of certain liquidity-support buybacks in old nominal securities within the ten-to-twenty-year and twenty-to-thirty-year sectors. Its stated purpose was to support the liquidity of older issues. It announced neither a yield ceiling nor the extinction of a claim held by the Fed. U.S. Treasury, 19 August 2026 release

At the moment Mélenchon spoke, that specific increase had not yet been executed. The broader buyback programme has existed since 2024, but the measure announced four days earlier was scheduled to begin seventeen days later.

How a Treasury buyback works

In a Treasury buyback, Treasury voluntarily repurchases an old bond from an investor at the price accepted through a competitive process. The delivered security is legally retired. But Treasury pays the seller with its cash balance, funded by taxes and, in a deficit period, mostly by other issuance. Treasury stated in 2024 that buybacks would not materially change net borrowing from the public because additional issuance replaces the repurchased securities. TreasuryDirect, buyback mechanics, U.S. Treasury, July 2024 borrowing estimate

The operation resembles replacing a collection of old, less-liquid bonds with newer securities that trade more easily. It can improve liquidity, smooth maturities or reduce some costs. It does not place debt outside time and does not spare Treasury from raising the required dollars.

The Fed is simultaneously conducting another operation. Its July 2026 Monetary Policy Report records nearly $250 billion in Treasury-bill purchases since the start of the year, including about $160 billion in reserve management purchases and $90 billion in reinvested MBS principal. Those are central-bank purchases of public debt, but the Fed expressly distinguishes them from QE. In August, the Desk planned about $17 billion of reinvestment purchases and no new reserve management purchases for the period from 14 August to 14 September. Federal Reserve, July 2026 Monetary Policy Report, New York Fed, operation schedule

Three institutions and three balance sheets must remain separate:

1. Treasury repurchases and retires old bonds while continuing to issue debt to finance the government;
2. the Fed buys bills in the secondary market against reserves to implement monetary policy;
3. the ECB and national central banks have bought public securities from twenty-one countries within a shared legal and monetary framework.

Mélenchon is right to recall that public balance sheets can be managed and that central banks hold some government debt. His comparison with the U.S. buybacks announced in August 2026 is nevertheless mistaken: those operations place no public debt “in the freezer”. They are not cancellation, a moratorium, QE or YCC.

Why the ECB cannot simply cancel France’s debt

The objection seems natural. The Eurosystem has bought French government bonds. The Banque de France belongs to the public sector. Why not delete the claim, note that the state partly owed money to itself and move on?

The starting point must be exact: the ECB and Banque de France can buy OATs in the secondary market, and they have done so under the PSPP and PEPP. The issue is not whether a French bond can appear on the Eurosystem balance sheet. It is the common objective, cross-country allocation, limits and exit strategy governing that holding. Buying whatever quantity France requested to maintain the yield chosen by its government would be a different policy.

The perimeter also needs correcting. The European Union has 27 states, but common monetary policy applies to the euro area, which has had 21 members since Bulgaria joined on 1 January 2026. The Eurosystem consists of the ECB and those 21 national central banks. French debt is not held in a private conversation between the finance ministry and a Banque de France free to decide on its own. ECB, Bulgaria’s entry

1. The current legal framework closes the door

Article 123 of the Treaty on the Functioning of the European Union prohibits the ECB and national central banks from providing credit facilities to public authorities or purchasing their debt directly. PSPP and PEPP purchases took place in the secondary market, with safeguards intended to prevent them from becoming the equivalent of direct financing.

The treaty does not contain the literal sentence “every cancellation is prohibited”. The ECB’s institutional position is nevertheless unambiguous. Christine Lagarde wrote to the European Parliament in 2021 that cancelling government debt held by the ECB would be incompatible with the treaties because it would amount to monetary financing. This is the institution’s official legal interpretation, not a specific Court of Justice ruling on an actual cancellation. Letter from the ECB President, 23 April 2021

Changing that rule would mean changing Europe’s framework. An ECB research publication in 2026 concludes that revising the treaty prohibition would require agreement by the countries and parliaments concerned. This would be a joint constitutional and fiscal choice, not an accounting entry adopted for France alone. ECB, Occasional Paper 397

2. The central-bank liability remains

When a central bank buys a French bond from a bank, its balance sheet gains an asset, the government security, and creates reserves as a liability. If it then cancels the bond, the asset disappears. The reserves held by the banking system do not automatically disappear. The loss reduces central-bank income, provisions or equity.

This does not mean a central bank is an ordinary company that must close when its equity becomes negative. It creates its currency and can operate through losses. It means the fiscal benefit is not free. Future income transferred to the Treasury falls, the capacity to absorb other losses is reduced, and a public recapitalisation could become necessary if credibility or financial independence required it. Banque de France’s 2025 management report illustrates how fixed-rate securities bought when yields were low coexist with reserves now remunerated at higher rates, which can suspend transfers to the state. Banque de France, 2025 management report

The ECB paper describes the counterpart: cancellation would generate a substantial loss and reduce central-bank distributions to governments. It also recalls that the securities were purchased from banks against remunerated reserves. ECB, Occasional Paper 397, sections 4 and 5

That mechanism also corrects the phrase “dilute the euro”. A bond purchase creates reserves used between banks. It does not automatically credit every household’s account or cause a proportional fall in the currency’s purchasing power. The effect on inflation and the exchange rate depends on the operation’s size, expected duration, the rate paid on reserves, the state of the economy, the credit response and confidence in the central bank.

Permanent, privileged financing for France could weaken the euro if investors concluded that price stability had become secondary or expected the same treatment for every other state. That depreciation would work through expectations and relative policy, not a mechanical euro-for-euro dilution. The most direct risk is institutional: subordinating a common mandate to one national budget’s solvency.

3. French cancellation would be a fiscal transfer inside a common currency

Under the PSPP, national central banks bought much of their own governments’ debt and some risks remained national. The ECB says 20% of purchases were subject to full risk sharing. That architecture was designed precisely to limit automatic fiscal transfers among countries. ECB, APP explainer

Cancelling only French securities would therefore create two problems at once. For portfolios held by Banque de France, France would also lose income and value inside its own central bank. For the mutualised share, other euro members would bear part of the loss. Either way, the operation would give one government a fiscal benefit that the other twenty could claim in turn.

The issue is not an abstract moral judgment about French or German virtue. It is an allocation rule. A common central bank cannot grant a permanent transfer to one national budget without deciding why others are ineligible and without changing expectations about future purchases.

4. Financing the next deficit

Assume, despite these obstacles, that the Eurosystem cancels some French bonds. The debt ratio falls immediately. But if revenue remains below spending before interest, France must issue new securities. Investors will price them according to the probability that the new debt will also be cancelled.

The result can be paradoxical: stock relief reduces the burden of the past while a higher risk premium raises the cost of future flows. The study by Cruces and Trebesch cannot calculate a French yield, but it documents the historical association between deeper haircuts, higher spreads and longer market exclusion.

5. Cancellation removes a monetary-policy instrument

Asset purchases are designed as a reversible tool. When inflation is too low, the Eurosystem buys securities. During normalisation, it can stop reinvestments, let bonds mature or, in principle, sell them. Permanent cancellation removes that option while the reserves created in the purchase remain in the system.

The debate then becomes one of fiscal dominance: does monetary policy pursue price stability or the solvency of one national budget? In a currency union without a federal budget comparable to that of a state, this boundary underpins political trust among members.

Karl Whelan’s academic analysis locates that boundary in European case law. It explains that OMT, PSPP and PEPP already create a strong interaction between fiscal solvency and monetary policy while remaining constrained by the prohibition of monetary financing and the Gauweiler and Weiss judgments. The working paper does not decide the legality of a hypothetical cancellation. It explains why holding limits and reversibility are economically and legally structural. Whelan, University College Dublin, 2022

Under the current framework, a common cancellation would require a major political and legal transformation. It would deliver real debt relief to states, but would also create a balance-sheet loss, transfers among members, a precedent for future debt and a constraint on monetary policy. It would amount to federal fiscal policy conducted through the central bank, with consequences well beyond an accounting entry.

Eight questions for examining a restructuring

The analysis of a specific case starts with these eight questions.

1. Is the debt genuinely unsustainable? A temporary liquidity crisis does not require the same response as a debt stock that cannot be stabilised.
2. Which debt is targeted? International bonds, local banks, bilateral creditors, multilateral institutions and suppliers have different contracts and social effects.
3. In which currency must it be paid? A state owing its own currency has more options than a dollarised country or one indebted in foreign currency.
4. Who holds the bonds? A foreign fund’s haircut does not travel like the loss of a local bank, insurer or pension fund.
5. Can the state finance its primary balance? After default, it must operate without asking the same creditors to fill the same hole immediately.
6. Is transition financing available? Reserves, exports, the IMF, bilateral partners or a regional mechanism determine whether imports and banks survive the negotiation. Our guide to reading an IMF programme separates financing, adjustment and sustainability.
7. Is restructuring deep and fast enough? Too shallow, it prepares a second default. Too late, it allows recession to grow. A brutal, conflictual process raises reputational and financing costs.
8. Who is explicitly protected? Deposits, small pensions, medicine, energy and social spending cannot be protected by declaration. They require financing inside the plan.

Sovereign CDS and spreads price market risk. They do not answer these distributional questions. Our guide to European sovereign debt separates the debt stock, maturity structure, holders and financing need.

Common shortcuts

The scenario of automatic apocalypse does not survive historical comparison. The state does not disappear, and genuinely unsustainable debt sometimes has to be restructured. Germany in 1953 and the HIPC initiatives indicate that well-designed relief can support growth and social expenditure.

Cancellation does not finance the deficit, banks, imports or investment. An extinguished claim also leaves a loss with its holder.

In Ecuador, the operation targeted two bonds while the country continued other payments. Oil, Chinese financing and a broader social policy also mattered. The fall in poverty is established; its attribution to default alone is not.

The German agreement does not validate unilateral refusal. It was a multilateral, geopolitical and financial restructuring designed to restore payment capacity and trade.

A bond held by the ECB has not disappeared economically. Some interest can return to states through central-bank profits, but the reserves created remain a remunerated liability and losses reduce distributions. In the Eurosystem, the balance sheet also serves as a sharing rule among twenty-one national budgets.

The choice concerns loss allocation and reconstruction

A country in distress does not choose between honouring every promise and making debt vanish. It chooses when to recognise the loss, how to allocate it and with which financing to rebuild afterwards.

Paying for too long can transfer resources to creditors while destroying the economy that was meant to reimburse them. Stopping abruptly can shatter banks, savings and access to foreign currency. Coordinated relief can free resources for social investment. Cancellation without reform or new financing can simply prepare the next default.

Ecuador, Germany, Argentina, Greece and Russia are not five versions of one story. They are five ways of locating a loss in an economy. The democratic question begins there: who chooses the bearer of the loss, who is protected during the transition, and what prevents the debt from returning?

Sources and method

This analysis relies first on texts and evaluations from the IMF, World Bank, ECB, Banque de France, Federal Reserve, U.S. Treasury, Bank of Japan, Reserve Bank of Australia, Bank of England, European Stability Mechanism, Bank of Greece, OECD, United Nations and national statistical institutes. Jean-Luc Mélenchon’s words are attributed to the full video of his 23 August rally with a timestamp, not to a press paraphrase. Ecuador’s 2009 and 2014 transactions are cross-checked against Reuters, Bloomberg and the IMF historical database. The academic corpus includes articles from the American Economic Review, Quarterly Journal of Economics, Journal of Finance, Journal of Money, Credit and Banking, Journal of the European Economic Association, Journal of International Economics, Journal of International Money and Finance, European Economic Review, European Review of Economic History and Industrial and Corporate Change. NBER and University College Dublin working papers are identified as such.

Limits

There is no observable counterfactual showing what Ecuador would have become without its 2008 default, Argentina without the 2001 missed payment or Greece with restructuring in 2010. Poverty comparisons use thresholds, surveys and methods specific to each period and should not be compared level for level across countries. Finally, the French cancellation scenario is not a forecast. It is used to make the current law and balance-sheet mechanics of the European Union and euro area explicit.